Accessibility settings

This paper is in the following e-collection/theme issue:

Applications of AI (343) Data Science (388) Artificial Intelligence (2341) Research Methodology - AI (48) Theoretical Innovations in AI (8) Foundations of AI (66)

Published on in Vol 5 (2026)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/83425, first published .
Fuzzy Logic Approaches for Causal Inference in Health Care: Systematic Review

Fuzzy Logic Approaches for Causal Inference in Health Care: Systematic Review

Fuzzy Logic Approaches for Causal Inference in Health Care: Systematic Review

Authors of this article:

Jaime Jamett1, 2 Author Orcid Image ;   Patricio Melendez2, 3 Author Orcid Image ;   Ximena Collao-Ferrada2, 4, 5 Author Orcid Image ;   Karina Cordero-Torres2, 6 Author Orcid Image ;   Alejandro Veloz2, 7 Author Orcid Image
Article Authors Cited by Tweetations Metrics

    1Office of the Vice President for Academic Affairs, Universidad de Valparaíso, Blanco 951, Valparaíso, Chile

    2PhD Program in Sciences and Engineering for Health, Universidad de Valparaíso, Valparaíso, Chile

    3Department of Maxillofacial Radiology, Faculty of Dentistry, Universidad Andres Bello, Viña del Mar, Chile

    4Department of Preclinical Sciences, School of Medicine, Universidad de Valparaíso, Valparaíso, Chile

    5Interdisciplinary Center for Biomedical Research and Engineering for Health (MEDING), Universidad de Valparaíso, Valparaíso, Chile

    6Interdisciplinary Center for Health Studies, Universidad de Valparaiso, Valparaiso, Chile

    7School of Biomedical Engineering, Universidad de Valparaíso, Valparaíso, Chile

    *all authors contributed equally

    Corresponding Author:

    Jaime Jamett, DDS


    Background: Fuzzy logic has been progressively investigated as a viable alternative to traditional statistical and machine learning methods in health care modeling, especially in environments marked by uncertainty, nonlinearity, and missing information. Although its use in prediction, classification, and risk stratification is well established, its application to explicit causal inference remains limited, varied, and methodologically premature.

    Objective: This systematic review aimed to examine how fuzzy logic frameworks have been used to address causal questions in health care, focusing on their methodological characteristics, comparative performance, and degree of integration with formal causal inference approaches.

    Methods: A systematic search across 6 databases (PubMed, Web of Science, ScienceDirect, SpringerLink, Scopus, and IEEE Xplore) identified peer-reviewed studies published between 2014 and 2025 that applied fuzzy modeling in health care settings with explicit or implicit causal objectives. The review adhered to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines and used a modified PICO (population, intervention, comparator, and outcome) framework for study selection. Data were extracted on the health care domain, fuzzy method, comparator use, and causal framing. Risk of bias was evaluated using the Joanna Briggs Institute (JBI) checklist and the PROBAST+AI tool, according to study design.

    Results: A total of 37 studies met the inclusion criteria. The most frequently applied approaches were fuzzy inference systems, fuzzy cognitive maps, and neuro-fuzzy models, with applications spanning infectious diseases, cancer, cardiovascular health, mental health, and occupational health. Fourteen studies included comparator models; among these, 5 reported superior performance of fuzzy approaches, 3 showed comparable results, and 6 lacked sufficient detail for a robust comparison. Only 2 studies explicitly implemented formal causal inference frameworks, while most relied on predictive or associative modeling with implicit causal assumptions. Overall, the risk of bias was moderate to high.

    Conclusions: Fuzzy logic offers interpretability and flexibility well suited to complex health care problems, yet its application to explicit causal inference remains fragmented. Greater methodological transparency, systematic benchmarking, and integration with formal causal designs—such as counterfactual and target trial frameworks—are required to establish fuzzy logic as a robust paradigm for causal inference in health care.

    Trial Registration: PROSPERO CRD420251044493; http://www.crd.york.ac.uk/PROSPERO/view/CRD420251044493

    JMIR AI 2026;5:e83425

    doi:10.2196/83425

    Keywords

    fuzzy logic; causality; delivery of health care; clinical decision-making; health information systems 


    Health care is strongly influenced by uncertainty. Clinical and public health decisions are routinely made under conditions of incomplete, ambiguous, or imprecise information, arising not only from individual patient variability but also from the complexity of health care systems and the processes through which real-world data are generated [1]. Such uncertainty encompasses both stochastic variability and epistemic constraints and is further amplified by heterogeneous, noisy, and nonlinear data from electronic health records, diagnostic imaging, physiological signals, and population-level monitoring systems [2]. Under these conditions, conventional statistical approaches—typically relying on fixed thresholds, linearity, and prespecified functional forms—often struggle to represent gradual transitions, ambiguous diagnostic boundaries, and context-dependent relationships that characterize real-world clinical data [1-3].

    Causal inference emerged as a methodological approach for estimating the effects of exposures or interventions on health outcomes, explicitly addressing the limitations of purely associational analyses [4-10]. Rather than focusing on prediction, this approach centers on counterfactual questions—what would be expected to occur under hypothetical alterations in treatment or exposure—by making causal assumptions explicit and, in principle, empirically assessable [9,11-13]. This perspective is particularly relevant in health care and public health, where randomized controlled trials are frequently impractical and observational data constitute the primary source of evidence [9,11,13,14]. In such settings, causal reasoning is commonly formalized using directed acyclic graphs, which encode assumptions about causal structure, confounding, and intervention pathways, thereby enabling principled identification of causal effects [15-17].

    Recent developments have emphasized the central role of explicit study design in strengthening causal inference from observational data. Target trial emulation (TTE) clarifies the causal question by prespecifying the key protocol components of a hypothetical randomized trial—including eligibility criteria, treatment strategies, time zero, follow-up, and outcomes—prior to analysis, thereby aligning observational studies with the core principles of randomized experiments [17-20]. While this design-oriented framework can reduce avoidable biases and enhance interpretability, it does not by itself ensure valid effect estimation. In practice, TTE still requires appropriate identification assumptions and estimation strategies and may remain vulnerable to challenges such as model misspecification or limited flexibility when representing complex data structures [17].

    Despite advances in causal inference and design-oriented approaches, substantial challenges persist at the estimation stage when analyzing complex health care data. Even when causal questions are explicitly defined, commonly used estimation methods often rely on inflexible functional assumptions, sharply delineated variables, and correctly specified models—conditions that are difficult to sustain in high-dimensional and heterogeneous clinical environments [3,8,11]. As a result, a methodological gap remains between rigorously specified causal designs and the flexible representation of nonlinearity, gradual clinical thresholds, and uncertainty inherent in observational health data, limiting the applicability of traditional causal models in complex real-world settings [1,2,11,14,21].

    As a response to the demand for flexible representations of uncertainty and nonlinearity, fuzzy logic has been adopted as a modeling paradigm in health care research, grounded in earlier theoretical developments on vagueness and graduality. Central to this evolution was Zadeh’s introduction of fuzzy sets as a generalization of classical set theory, in which membership is defined by degrees rather than binary inclusion [22-24]. This formulation provided a formal mathematical basis for representing ambiguity, partial truth, and gradual transitions in complex systems, thereby enabling the representation of phenomena that cannot be adequately captured using crisp categories. Building on this foundation, subsequent developments extended fuzzy sets into operational fuzzy logic systems, particularly through rule-based inference mechanisms that support reasoning with linguistic variables and imprecise conditions [25].

    In clinical contexts, this representational flexibility facilitates the translation of gradual and linguistically defined clinical concepts into implementable computational models. Building on these foundations, a range of fuzzy logic–based approaches—including fuzzy inference systems (FIS), adaptive neuro-fuzzy inference systems (ANFIS), fuzzy cognitive maps (FCM), Takagi-Sugeno models, and fuzzy clustering—have been applied across diverse health care domains. These applications span infectious diseases [26-34], cardiology [35-42], oncology [43-49], obstetrics [50-52], mental health [53,54], and occupational health and safety [55-59].

    More recently, fuzzy logic–based approaches have increasingly been combined with machine learning and artificial intelligence techniques to enhance predictive performance, scalability, and automation in health care applications [60,61]. Despite this growing convergence, the extent to which such hybrid models explicitly engage with causal reasoning—through the definition of counterfactual estimands, formal identification strategies, and transparent causal assumptions—remains inconsistently reported in the literature. Against this background, this systematic review aimed to evaluate and synthesize evidence on the application of fuzzy logic–based approaches for causal inference in health care.


    Design and Reporting Standards

    The review was conducted following established systematic review standards, with adaptations to accommodate computational health modeling studies [62]. Eligibility criteria and search strategy used a modified PICO framework, targeting complex datasets (Population), fuzzy logic for causal inference (Intervention), conventional modeling methods (Comparator), and performance or interpretability outcomes (Outcomes).

    The review was conducted and reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines to ensure transparency and methodological rigor [63]. A systematic search was performed across 6 bibliographic databases, with references managed using Zotero (v7.0.15) and blinded title-abstract and full-text screening conducted in Rayyan (Qatar Computing Research Institute). The review protocol was prospectively registered in PROSPERO (registration number CRD420251044493). Risk of bias was assessed using the Joanna Briggs Institute (JBI) checklist for analytical cross-sectional studies [64] and the PROBAST+AI tool for predictive modeling studies [65]. The procedures applied at each stage of the review are described in detail below.

    This systematic review addresses a critical gap in the literature regarding how fuzzy logic–based approaches have been used to support causal inference in health care. The primary research question guiding the review was: How have fuzzy modeling approaches been applied, alone or in combination with other methods, to address causal questions in complex, multivariable health care settings? Rather than testing superiority, the review aimed to examine the contexts, modeling strategies, and assumptions under which fuzzy logic–based methods have been used in relation to causal objectives, particularly in settings characterized by uncertainty, nonlinearity, and high-dimensional data.

    Research Questions and Scope

    To structure this analysis, three secondary questions were defined: (RQ1) What modeling characteristics and design features are commonly reported in fuzzy-based approaches addressing causal questions? (RQ2) Under what data or problem contexts are fuzzy logic–based methods compared with conventional modeling approaches? (RQ3) How are the resulting insights framed in relation to clinical or policy-relevant decisions?

    Eligibility Criteria

    Studies were eligible for inclusion if they applied fuzzy logic–based approaches in health care settings and demonstrated either an explicit or implicit causal objective. Explicit causal intent was defined using formal causal frameworks, counterfactual reasoning, or clearly articulated intervention contrasts. Implicit causal intent was identified when modeling structures, analytical interpretations, or conclusions were framed in terms of causal effects, intervention impact, or decision-relevant implications beyond prediction. This inclusive criterion allowed the review to capture both formally specified and informally articulated causal approaches.

    Studies were excluded if they did not use fuzzy-based techniques, lacked any causal objective, or focused exclusively on diagnostic classification or prediction without causal interpretation.

    Information Sources and Search Strategy

    The literature search was conducted between March and June 2025 across 6 electronic databases: PubMed, Web of Science, ScienceDirect, SpringerLink, Scopus, and IEEE Xplore. Search strategies combined controlled vocabulary terms (eg, MeSH [Medical Subject Headings] in PubMed) with platform-specific free-text keywords to capture studies at the intersection of fuzzy logic, causal inference, and health care. To enhance sensitivity, the search was intentionally broad and complemented by manual screening of reference lists from included studies. The full search strategies for each database are reported in Table 1.

    Table 1. Search strategies across databases for the identification of relevant studies.
    DatabaseSearch strategy
    PubMed(“Fuzzy Logic” [MeSH] OR “Fuzzy logic” [Title/Abstract] OR “Fuzzy modelling” [Title/Abstract] OR “Fuzzy inference system*” [Title/Abstract])
    AND
    (“Causality” [MeSH] OR “Causal Inference” [Title/Abstract] OR “Causal Model*” [Title/Abstract] OR “Causal Discovery” [Title/Abstract])
    AND
    (“Healthcare” [Title/Abstract] OR “Medical Informatics” [MeSH] OR “Clinical Decision-Making” [MeSH] OR health* [Title/Abstract] OR clinical [Title/Abstract] OR medical [Title/Abstract])
    Web of Science, ScienceDirect, Springer, IEE Xplore, and Scopus(“fuzzy logic” OR “fuzzy modelling” OR “fuzzy inference system*”)
    AND
    (“causal inference” OR “causal model*” OR “causal discovery”)
    AND
    (health* OR clinical OR medical OR “medical informatics” OR “clinical decision*”)

    Study Selection Process

    Eligible studies were required to be peer-reviewed, published in English between 2014 and 2025, and to provide sufficient methodological detail to allow critical appraisal. Only original research articles with accessible full text and direct relevance to clinical or health policy decision-making were included. While the reporting of performance metrics (eg, accuracy and area under the curve, AUC) and the use of comparator methods were encouraged, their absence did not constitute grounds for exclusion when studies provided substantive contributions to fuzzy modeling or causal reasoning in health care.

    Studies were excluded if they were not written in English, did not constitute original research (including narrative or systematic reviews, editorials, commentaries, or conference abstracts without full text), were published outside the predefined time frame, involved extremely small samples (fewer than 5 observations), or lacked sufficient methodological transparency to support reproducibility or critical appraisal. These criteria were applied to ensure inclusion of studies with empirical relevance, conceptual rigor, and clarity in reporting.

    Study selection was managed using Zotero and Rayyan (Qatar Computing Research Institute). Two reviewers (JJ and KC-T) independently screened titles, abstracts, and full texts according to predefined inclusion and exclusion criteria. Discrepancies were resolved through discussion with a third reviewer (PM). Interreviewer agreement was 94%, and final inclusion decisions were reached by consensus, with oversight provided by additional authors (XC-F and AV).

    Data Extraction and Classification

    After removal of duplicates (n=6) and clearly irrelevant records (n=390), 225 records were retained for title and abstract screening. Of these, 153 were excluded based on predefined inclusion criteria, leaving 72 full-text articles assessed for eligibility. No reports were lost during retrieval. Thirty-five full-text articles were excluded, most commonly due to publication outside the predefined time frame (n=27), as well as wrong population (n=2), wrong outcome (n=3), wrong publication type (n=1), or wrong study design (n=2). A total of 37 studies were included in the final synthesis.

    Following the inclusion of 37 studies, a structured data extraction process was implemented to ensure consistency while accommodating methodological heterogeneity. Two reviewers independently extracted data using a piloted extraction form, with discrepancies resolved through consensus or, when necessary, third-party adjudication. The extraction framework was designed to capture both technical modeling features and elements relevant to causal framing and interpretation.

    Evidence Synthesis

    Extracted variables were organized across four domains: (1) bibliographic and contextual information (author, year, journal, and health care domain); (2) data characteristics (data source, dataset type, and sample size); (3) modeling and analytical features, including fuzzy modeling framework (eg, FIS, FCM, neuro-fuzzy, and Takagi-Sugeno), comparator methods (eg, generalized linear models, structural equation models, and directed acyclic graph–informed analyses), and reported performance metrics (eg, accuracy, AUC, and root mean square error); and (4) elements related to causal framing, including stated causal assumptions, interpretability features, and reported clinical or policy implications.

    To support consistency across studies and reduce terminological heterogeneity, ELICIT—an artificial intelligence–assisted evidence synthesis platform—was used to standardize extracted terminology, assist in the classification of modeling approaches, and check internal coherence of extracted items. ELICIT was used as a supportive tool for data organization and synthesis and did not replace reviewer judgment in data extraction or interpretation.

    Risk of Bias Assessment

    Risk of bias was assessed using two complementary tools, selected according to the methodological design of each included study. The JBI checklist [64] was applied to studies with observational or associational designs, particularly those using structural causal reasoning without formal identification strategies. Studies focused on predictive model development or validation were evaluated using the PROBAST+AI tool [65].

    Tool-specific criteria guided the assessment of potential bias. For studies evaluated with PROBAST+AI, emphasis was placed on outcome definition, predictor handling, and analytical transparency. For studies assessed using the JBI checklist, particular attention was given to reporting adequacy, conceptual rigor, and overall methodological clarity. These assessments identified recurrent limitations related to internal validity and reporting quality across the included evidence.

    Overall risk of bias was classified as low, moderate, or high based on the severity and frequency of methodological concerns identified using each assessment tool. Given the substantial methodological heterogeneity of the included studies, a formal GRADE (Grading of Recommendations Assessment, Development, and Evaluation) assessment was not performed. Instead, the certainty of the evidence was appraised qualitatively by triangulating risk of bias assessments, methodological coherence, and robustness of reporting.

    Data Synthesis and Analytical Strategy

    Following data extraction, findings were synthesized to characterize modeling approaches, identify comparative trends, and highlight evidence gaps at the intersection of fuzzy logic and causal inference. Descriptive statistics were used to summarize the distribution of included studies by health care domain, modeling approach, data source, and sample size, with frequencies and proportions calculated for fuzzy modeling techniques, data types (real-world or synthetic), and reported performance metrics.

    In parallel, a thematic analysis was conducted across 4 focal areas: diversity of modeling frameworks, comparative performance, treatment of causal assumptions, and relevance to clinical or policy decision-making. This analytical phase also identified recurring methodological limitations, reporting inconsistencies, and underexplored applications, thereby complementing the structured risk of bias assessments. Summary tables were used to support structured comparison across studies and ensure consistent classification by health care domain, fuzzy modeling technique, comparator method, and reported performance metrics.


    During the systematic search conducted between March and June 2025, a total of 621 records were identified across 6 electronic databases. After removal of duplicates and screening of titles, abstracts, and full texts, 37 studies published between 2014 and 2025 met the inclusion criteria and were retained for final synthesis. The PRISMA 2020 flow diagram (Figure 1) details the study selection process, including the number of records screened, excluded, and included at each stage of the review.

    Table 2 lists fuzzy logic–based methodologies from 37 studies, showing the literature’s methodological diversity. The table defines each approach, its main analytical role, common application in reviewed studies, and frequency of use.

    Table 3 summarizes the 37 studies included in the final synthesis, spanning health care domains such as infectious diseases, cardiovascular conditions, cancer, mental health, occupational health, and preterm birth. Across studies, the most frequently applied fuzzy approaches were FIS (Mamdani type), ANFIS, fuzzy analytic hierarchy process (FAHP) variants, and FCM. Sample sizes varied widely, ranging from fewer than 100 participants to large-scale public datasets exceeding 1000 cases, with data sources including institutional or hospital records, expert-based judgments, and simulated data.

    Fourteen studies used direct comparator methods, most commonly logistic regression, decision trees, or ensemble classifiers, whereas the remaining studies relied on baseline comparisons or did not include external benchmarks. Performance was typically reported using accuracy or AUC, with sensitivity and specificity included in selected cases. Importantly, only a minority of studies explicitly framed their analyses within formal causal inference paradigms, while most remained primarily predictive or associative in scope.

    The temporal distribution of the included studies shows an uneven pattern over the past decade, with episodic increases rather than a steady growth trajectory, culminating in a pronounced peak in 2024 (Figure 2).

    ‎
    Figure 1. Flow diagram illustrating the identification, screening, eligibility assessment, and inclusion of studies in the systematic review, according to PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) 2020 guidelines [63].
    Table 2. Fuzzy logic–based methods identified across the included studies (n=37)a.
    Fuzzy methodAbbreviationPrimary analytical roleTypical application in reviewed studiesStudies, n (%)
    Fuzzy inference system (Mamdani-type and variants)FISRule-based modeling under uncertaintyPrediction, classification, decision support8 (21.62)
    Fuzzy analytic hierarchy processFAHPMulticriteria decision analysisRisk prioritization, decision support6 (16.22)
    Fuzzy cognitive mapsFCMConceptual modeling of interacting variablesSimulation of influence structures, exploratory causal reasoning5 (13.51)
    Adaptive neuro-fuzzy inference systemANFISHybrid learning and fuzzy inferencePrediction and pattern recognition3 (8.11)
    Hybrid fuzzy models combined with MCDMbHybrid Fuzzy + MCDMMulticriteria decision supportStructural prioritization and ranking3 (8.11)
    Fuzzy clustering (C-means/K-means)—cUnsupervised pattern discoveryGrouping and exploratory analysis2 (5.41)
    Fuzzy-trace theoryFTTCognitive decision modelingBehavioral and decision-making analysis2 (5.41)
    Fuzzy-set qualitative comparative analysisfsQCAConfigurational causal analysisIdentification of necessary and sufficient conditions2 (5.41)
    Takagi-Sugeno fuzzy modelsTS/TSKRule-based functional approximationPredictive modeling1 (2.70)
    Fuzzy failure mode and effects analysisF-MEARisk and failure assessmentSafety and risk analysis1 (2.70)
    Mediative fuzzy logicMFLDecision mediation modelingClinical decision support1 (2.70)
    Fuzzy evidential reasoningFEREvidence aggregationDecision support under uncertainty1 (2.70)
    Likelihood-fuzzy analysisLFAProbabilistic-fuzzy integrationRisk estimation1 (2.70)
    Profile-based fuzzy association rule miningPB-FARMPattern and rule discoveryAssociation analysis1 (2.70)

    aSome categories in this table group closely related fuzzy logic–based methods that are reported under different specific names across individual studies (Tables 3 and 4). In particular, hybrid fuzzy models combined with multicriteria decision-making encompass approaches that integrate fuzzy rule–based systems with decision-analytic or influence-structuring techniques, such as decision-making trial and evaluation laboratory, analytic network process, analytic hierarchy process variants, or type-2 fuzzy sets. Similarly, several variants of fuzzy cognitive maps, neuro-fuzzy systems, and Takagi-Sugeno models are reported using study-specific nomenclature and are therefore grouped according to their underlying modeling principles. These groupings were applied to facilitate synthesis and comparability across studies with conceptually similar analytical objectives, while detailed methodological implementations, comparators, and performance metrics for each study are reported in Tables 3 and 4.

    bMCDM: multicriteria decision-making.

    cNot applicable.

    Table 3. Summary of characteristics and comparative outcomes of included studies.
    Study (year)DomainaTask/outcomebFuzzy methodcDataset sizedDataeComparatorfPrimary metricgCIh
    Amirkhani et al (2014) [66]OtherAutoimmune hepatitisNFCM+NFISMInstDirect: ANFISAUC 89.8I
    Lee et al (2015) [30]IDHIV prevalence (policy)fsQCALPubNoneConsistency 0.95E
    Maranate et al (2015) [67]OtherOSA severityFAHPLInstNoneSens 92.3P
    Subramanian et al (2015) [43]CancerBreast cancer riskL2-FCMSSynthDirect: FCMAUC 94.3P
    Wolfe et al (2015) [49]CancerRisk decision-makingFTTMPubDirect: RCT controlNRP
    Mollalo and Khodabandehloo (2016) [31]IDLeishmaniasis risk mapFAHP+GISLInstBaseAUC 90.5P
    Yılmaz et al (2016) [45]CancerLung cancerANFIS-MEPLInstDirect: ANFIS, EPAUC 94.6P
    Pota et al (2017) [47]CancerRadiotherapy side effectsLFASInstDirect: NBAUC 0.81P
    Stanković and Stanković (2017) [48]CancerProstate survivalNeuro-fuzzySInstDirect: ANN, FISR2=0.83P
    Iancu (2018) [42]CVDCVD diagnosisMFLNASynthNoneNRP
    Sabahi (2018) [39]CVDCHD risk rankingBFAHPNAExpDirect: AHPAUC 0.86P
    Saleh et al (2018) [41]CVDDiabetic retinopathyANFISMInstDirect: RF, MLP, kNNAUC 0.84P
    Argyropoulos et al (2019) [68]OtherAKI stage-3 riskTSKLInstDirect: LRAUC 0.95P
    Romero et al (2019) [33]IDDengue riskFIS-MamdaniNAPubNoneAUC >0.86P
    Sarkar et al (2019) [26]IDMalaria ecological riskFIS+AHPLMixedBaseNRP
    Souza et al (2019) [51]PTBPTB phenotypesFuzzy clusteringLInstNoneNRP
    Boni et al (2020) [35]CVDCVD in dialysisFIS-MamdaniMInstNoneAUC 0.92P
    Hynek et al (2020) [54]MentalRefugee mental healthFCMSExpNoneNRI
    Mahmoodi et al (2020) [44]CancerGastric cancerFCM-NHLMInstDirect: ANN, SVM, DT, NBAUC 95.8I
    Piyatilake and Perera (2020) [32]IDDengue clustersFAHPLPubBaseAUC 0.73P
    Malakoutikhah et al (2021) [58]OHSMSD risk (steel)FIS-MamdaniMMixedNoner=0.24P
    Shi et al (2021) [34]IDOutbreak riskFERSExpNoneα=0.79P
    Yavari et al (2021) [40]CVDHeart disease profilingPB-FARMMPubDirectConf 0.73P
    Mohandes et al (2022) [57]OHSConstruction safetyIVIF-DEMATEL+ANPSPubDirectα=0.74E
    Safaei et al (2022) [38]CVDObesity modelMFRBS+DEMATELLPubNoneNRI
    Barbounaki and Sarantaki (2022) [50]PTBPTB risk assessmentFAHPMInstBaseNRP
    Brust-Renck and Reyna (2023) [46]CancerCancer risk decisionsFTTLInstBaseNRP
    Aydın and Özkan (2024) [36]CVDLMIC cardiovascular risk profilingIVPF-AHP+TOPSISLInstNoneNRP
    Benito et al (2024) [28]IDCOVID/dengueFCM+LAMDALPubDirect: RF, LAMDAAUC 0.89I
    Chen et al (2024) [53]MentalChild depressionfsQCA+OLSMMixedNoneConsistency 0.867I
    Costa et al (2024) [29]IDLeishmaniasis riskFIS-MamdaniLPubNoneNRP
    Sakinala et al (2024) [56]OHSMining MSD riskFIS-MamdaniSInstBaseP=.19P
    Sümbül-Şekerci et al (2024) [69]OtherT2DM cognitionFCM+CRTMInstDirect: CRTAUC 0.91P
    Upadhyay et al (2024) [55]OHSIron ore MSD riskFIS-MamdaniSInstNoneNRP
    Demir and Sabır (2025) [59]OHSWorkplace riskF-FMEASExpDirect: FMEANRP
    Rani and Dhanasekar (2025) [27]IDZika risk factorsType-2 FS+MCDMNAExpNoner>0.92P
    Scrobota et al (2025) [70]IDPeriodontitis (T2DM)FIS-MamdaniSInstNoneP=.02P

    aID denotes infectious diseases; CVD, cardiovascular diseases; OHS, occupational health and safety; and PTB, preterm birth.

    bTask/outcome: OSA: obstructive sleep apnea; CVD: cardiovascular diseases; CHD: coronary heart disease; AKI: acute kidney injury; PTB: preterm birth; MSD: musculoskeletal disorders; LMIC: low- and middle-income countries; T2DM: type 2 diabetes.

    cFuzzy modeling approaches include fuzzy inference systems (FIS), adaptive neuro-fuzzy inference systems (ANFIS), fuzzy cognitive maps (FCM), fuzzy analytic hierarchy process (FAHP), fuzzy-set qualitative comparative analysis (fsQCA), Takagi-Sugeno-Kang models (TSK), fuzzy evidential reasoning (FER), fuzzy-trace theory (FTT), mediative fuzzy logic (MFL), and related hybrid extensions. neuro-fuzzy cognitive map (NFCM); geographic information system (GIS); t-norm modified Einstein operator (MEP); likelihood-fuzzy analysis (LFA); mediative fuzzy logic (MFL); bimodal fuzzy analytic hierarchy process (BFAHP); nonlinear Hebbian learning (NHL); profile-based fuzzy association rule mining algorithm (PB-FARM); interval-valued intuitionistic fuzzy decision-making trial and evaluation laboratory (IVIF-DEMATEL); Mamdani fuzzy rule-based system integrated with the decision-making trial and evaluation laboratory method (MFRBS-DEMATEL); interval-valued Pythagorean fuzzy analytic hierarchy process combined with the technique for order preference by similarity to ideal solution (IVPF-AHP-TOPSIS); fuzzy cognitive maps combined with learning algorithm for multivariate data analysis (FCM+LAMDA); fuzzy-set qualitative comparative analysis combined with ordinary least squares regression (fsQCA+OLS); fuzzy cognitive maps combined with classification and regression trees (FCM+CRT), fuzzy failure mode and effects analysis (F-MEA).

    dDataset size. S: small; M: medium; L: large; NA: not available.

    eData sources are classified as public (Pub), institutional or hospital-based (Inst), expert-based (Exp), synthetic (Synth), or mixed.

    fComparator methods are categorized as none (no comparator), baseline (base), or direct comparison with other models (direct). Abbreviations: analytic hierarchy process (AHP); adaptive neuro-fuzzy inference system (ANFIS); artificial neural network (ANN); classification and regression tree (CRT); decision tree (DT); evolutionary programming (EP); fuzzy cognitive map (FCM); fuzzy inference system (FIS); failure mode and effects analysis (FMEA); k-nearest neighbors (kNN); learning algorithm for multivariate data analysis (LAMDA); logistic regression (LR); multilayer perceptron (MLP); naive Bayes (NB); randomized controlled trial (RCT); random forest (RF); support vector machine (SVM).

    gPrimary performance metrics are reported as in the original studies and include area under the curve (AUC), accuracy, sensitivity, correlation coefficients, and reliability indices. Abbreviations: Conf: confidence (association rule confidence); NR: not reported.

    hCausal intent (CI) was classified as explicit (E) when formal causal inference frameworks were applied, implicit (I) when causal assumptions were suggested but not formally specified, and predictive/associative (P) when analyses focused on prediction without explicit causal interpretation.

    Table 4. Comparative evaluations of fuzzy logic models versus conventional statistical or machine learning approaches (n=14 studies). The table summarizes studies that directly compared fuzzy-based methods with traditional models, reporting predictive performance metrics and qualitative comparative outcomes.
    Study (year)DomainFuzzy methodComparatorsReported metricsComparative outcome
    Amirkhani et al (2014) [66]Other (AIHa)NFCMb+NFIScNFIS, ANFISd, HyFISeAccf 89.8Neuro-fuzzy cognitive map improved explainability; performance comparable
    Subramanian et al (2015) [43]Cancer (breast)L2-FCMgStandard FCMhAcc 94.3 vs 92.6Layered FCM improved accuracy and interpretability
    Yılmaz et al (2016) [45]Cancer (lung)ANFIS-MEPiANFISAcc 94.6 vs 92.6; RMSEj lowerNeuro-fuzzy model achieved higher accuracy and faster convergence
    Pota et al (2017) [47]Cancer (RTk side effects)LFAlNaïve BayesAcc 0.81 vs 0.84; mixed sensitivity/specificityComparable accuracy: fuzzy model offered rule-based interpretability
    Stanković and Stanković (2017) [48]Cancer (prostate)Neuro-fuzzyANNm, FISnR2=0.83; RMSE lowestNeuro-fuzzy slightly outperformed ANN and standard FIS
    Saleh et al (2018) [41]CVDo (DRp)ANFISRFq, MLPr, kNNs, DRSAtAcc 84.2 vs 77.3 (DRSA)ANFIS achieved the best accuracy among the tested classifiers
    Sabahi (2018) [39]CVD (CHDu risk)BFAHPvAHPwAcc 85.9 vs 77.3Fuzzy AHP showed greater robustness under uncertainty
    Argyropoulos et al (2019) [68]AKIxTSKyLogistic regressionAUCz 0.95 vs 0.95Equivalent AUC; fuzzy gained sensitivity in some models
    Mahmoodi et al (2020) [44]Cancer (gastric)FCM-NHLaaANN, SVMab, DTac, NBadAcc 95.8 vs 90.5 (ANN)FCM-NHL achieved the highest predictive accuracy across methods
    Yavari et al (2021) [40]CVD (heart disease)PB-FARMaeAssociation rule/classification methodsSupport/confidenceFuzzy association mining extracted higher-confidence clinical rules
    Mohandes et al (2022) [57]OHSaf (safety)IVIF-DEMATELag+ANPahIVIF-ANPaiReliability α=0.74Hybrid fuzzy method prioritized causal factors with higher consistency
    Benito et al (2024) [28]IDaj (COVID/dengue)FCM+LAMDAakRFAUC 0.89 vs 0.98 (RF)RF outperformed in accuracy; fuzzy models offered stronger explainability
    Sümbül-Şekerci et al (2024) [69]Other (T2DMal cognition)FCM+CRTamCRTAUC 0.91 (cluster 1)Fuzzy clustering identified cognitive subgroups; CRT supported classification
    Demir and Sabır (2025) [59]OHSF-FMEAanClassical FMEAaoNRapFuzzy FMEA reduced subjectivity in risk prioritization

    aAIH: autoimmune hepatitis.

    bNFCM: neuro-fuzzy cognitive map.

    cNFIS: neuro-fuzzy inference system.

    dANFIS: adaptive neuro-fuzzy inference systems.

    eHyFIS: hybrid fuzzy inference system.

    fAcc: accuracy.

    gL2-FCM: layered fuzzy cognitive map.

    hFCM: fuzzy cognitive map.

    iMEP: t-norm modified Einstein operator.

    jRMSE: root mean square error.

    kRT: radiation therapy.

    lLFA: likelihood-fuzzy analysis.

    mANN: artificial neural network.

    nFIS: fuzzy inference system.

    oCVD: cardiovascular disease.

    pDR: diabetic retinopathy.

    qRF: random forest.

    rMLP: multilayer perceptron.

    skNN: k-nearest neighbors.

    tDRSA: dominance-based rough set approach.

    uCHD: coronary heart disease.

    vBFAHP: bimodal fuzzy analytic hierarchy process.

    wAHP: analytic hierarchy process.

    xAKI: acute kidney injury.

    yTSK: Takagi-Sugeno-Kang.

    zAUC: area under the curve.

    aaNHL: nonlinear Hebbian learning.

    abSVM: support vector machine.

    acDT: decision tree.

    adNB: naïve Bayes.

    aePB-FARM: profile-based fuzzy association rule mining.

    afOHS: occupational health and safety.

    agIVIF-DEMATEL: interval-valued intuitionistic fuzzy decision-making trial and evaluation laboratory.

    ahANP: analytic network process.

    aiIVIF-ANP: interval-valued intuitionistic fuzzy analytic network process.

    ajID: infectious diseases.

    akLAMDA: learning algorithm for multivariate data analysis.

    alT2DM: type 2 diabetes mellitus.

    amCRT: classification and regression tree.

    anF-FMEA: fuzzy failure mode and effects analysis.

    aoFMEA: failure mode and effects analysis.

    apNR: not reported.

    ‎
    Figure 2. Chronological distribution of included investigations (2015‐2025). The bar chart illustrates the annual number of studies published throughout the review period.

    The most frequently addressed conditions were infectious diseases (n=10) [26-34,70], cardiovascular diseases (n=7) [35,36,38-42], cancer (n=7) [43-49], occupational health and safety (n=5) [55-59], mental health (n=2) [53,54], and preterm birth (n=2) [50,51]. Additional studies fell into miscellaneous categories [66,68,69].

    Regarding data sources, most studies used institutional or hospital datasets (n=18) [31,35,36,41,44-48,50,51,55,56,66-71], while 9 relied on public datasets [28-30,32,33,38,40,49,57], 5 reported expert-based data [27,34,39,54,59], 3 reported mixed sources [26,53,58], and 2 used simulated or synthetic data [42,43].

    Sample sizes varied considerably across studies: 13 used large datasets (n≥1000) [26,28-32,36,38,45,46,51,67,68], another 10 relied on medium-sized samples (n=100 to 999) [35,40,41,44,49,50,53,58,66,69], 10 used small datasets (n<100) [34,43,47,48,54-57,59,68,70], and in 4 studies, the sample size was not applicable [27,33,39,42].

    A broad array of fuzzy logic techniques was identified across the included studies, reflecting substantial methodological heterogeneity. The most commonly used methods were FIS and their variations (n=8) [26,29,33,35,55,56,58,70], frequently implemented using Mamdani-type structures; followed by the FAHP (n=6) [26,31,32,39,50,67], FCM (n=5) [28,43,44,54,66,69], adaptive neuro-fuzzy systems (n=3) [41,45,48], typically combining neural architectures with fuzzy rule bases for improved learning capacity and hybrid fuzzy approach combined with multicriteria decision models (n=3) [27,38,57].

    Other used models were fuzzy clustering (C or K means) (n=2) [51,67], fuzzy-trace theory (n=2) [46,49], fuzzy-set qualitative comparative analysis (fsQCA) (n=2) [30,53], Takagi-Sugeno models (n=1) [66], fuzzy failure mode and effects analysis (n=1) [59], mediative fuzzy logic (n=1) [42], fuzzy evidential reasoning (n=1) [34], likelihood-fuzzy analysis (n=1) [47], and profile-based fuzzy association rule mining (n=1) [40].

    Among the studies reviewed, 14 conducted direct comparative evaluations against traditional methods such as logistic regression, decision trees, or standard statistical models [28,39-41,43-45,47,48,57,59,66,68,69] and 6 studies used baseline comparisons, typically involving simple pre/post assessments without an external benchmark [26,31,32,46,50,56]. In contrast, 17 studies applied fuzzy modeling in isolation, without any form of benchmarking or comparator method, relying solely on internal outputs to assess performance [27,29,30,33-36,38,42,49,51,53-55,58,67,70]

    Between the studies that conducted direct comparative evaluations, 5 reported that fuzzy logic models outperformed traditional methods, including statistical classifiers and machine learning algorithms. These included Mahmoodi et al [44], who achieved 95.8% accuracy in gastric cancer prediction using FCM; Yılmaz et al [45], who obtained 94.6% accuracy with a neuro-fuzzy model for lung cancer; Subramanian et al [43], who reported 94.3% overall accuracy using a layered FCM for breast cancer risk; Sabahi [39], who introduced a bimodal FAHP model with accuracies above 85%; and Saleh et al [41], whose ANFIS classifier outperformed other ensemble models in diabetic retinopathy detection.

    Three studies showed that fuzzy models yielded comparable or slightly superior performance relative to conventional methods. Argyropoulos et al [68] reported equivalent AUC values for both fuzzy logic and logistic regression models in predicting acute kidney injury, while Pota et al [47] found similar predictive accuracy between likelihood-fuzzy analysis and naïve Bayes classifiers in radiotherapy toxicity. Stanković and Stanković [48] also demonstrated that a neuro-fuzzy system marginally outperformed an artificial neural network in predicting prostate cancer survival.

    The remaining 6 studies—Amirkhani et al [66], Yavari et al [40], Mohandes et al [57], Benito et al [28], Sümbül-Şekerci et al [69], and Demir and Sabır [59]—involved direct comparisons but did not report sufficient methodological or statistical detail to clearly assess the relative effectiveness of the fuzzy approach. To visually summarize the comparative performance of fuzzy logic models versus conventional statistical approaches, Figure 3 presents reported accuracy values from studies that provided quantifiable metrics. Only those studies with explicit accuracy comparisons were included, enabling a focused assessment of relative predictive performance across diverse health care domains.

    Across these studies, common performance metrics included accuracy (84%‐95.8%), AUC (0.70‐0.95), and error measures such as root mean square error, mean absolute error, and mean squared error. These results underscore the adaptability of fuzzy modeling to clinical decision-making contexts marked by uncertainty, incomplete data, and the need for interpretability.

    In terms of causal inference, conceptual approaches varied across the studies. While most of the studies addressed high-complexity settings involving multiple interacting variables, only two explicitly adopted formal causal inference frameworks. These included Lee et al [30], who used fsQCA with sufficiency and necessity thresholds; and Mohandes et al [57], who implemented a hybrid interval-valued intuitionistic fuzzy DEMATEL-ANP (decision-making trial and evaluation laboratory analytic network process) model with cross-validation.

    Six additional studies [28,38,44,53,54,66] simulated causal mechanisms using methods such as iterative expert-based system mapping or FCM. However, none of these studies explicitly operationalized a formal causal inference framework grounded in counterfactual reasoning or directed acyclic graphs. Instead, causal assumptions were inferred through expert consensus or embedded in the structure of fuzzy systems.

    The remaining 29 studies used fuzzy logic primarily for predictive or associative analysis [26,27,29,31-36,39-43,45-51,55,56,58,59,67-70], with causal relationships often left implicit, untested, or loosely derived from domain-specific knowledge alone.

    Table 3 provides a detailed synthesis of the 14 studies that directly compare fuzzy logic models with traditional statistical or machine learning methods. Most of these studies reported performance gains for fuzzy approaches, particularly in cancer [43-45] and cardiovascular domains [39,41]. In several cases, fuzzy models offered not only higher accuracy or sensitivity but also enhanced interpretability. Others showed broadly comparable results with added value in robustness [47,48,68]. A smaller group reported either mixed outcomes or limited statistical detail, emphasizing interpretability and methodological novelty over raw predictive gains [28,40,57,59,66,69].

    ‎
    Figure 3. Reported accuracy values from studies performing direct quantitative comparisons between fuzzy logic–based models and conventional approaches [39,41,43-45,47,68].

    Collectively, the evidence summarized in Table 4 indicates that fuzzy logic–based approaches have been evaluated against conventional methods in a limited subset of studies, yielding heterogeneous results and variable reporting quality. Although several comparative assessments suggest potential advantages in managing uncertainty and enhancing interpretability, the absence of systematic benchmarking and the predominance of predictive objectives preclude definitive conclusions regarding comparative effectiveness.

    While the findings indicate that fuzzy logic–based approaches are frequently applied in predictive health care modeling, the strength of the available evidence must be interpreted considering methodological quality and risk of bias. Seventeen studies were evaluated using the PROBAST+AI tool, specifically designed for assessing bias in prediction model studies [65]. Of these, 9 were rated as having a high risk of bias [27,34,35,39,43,44,47,66,67], and 8 studies were rated as moderate risk [28,33,40,41,45,48,68,69], standing out for more robust validation procedures, detailed variable handling, and partial transparency. None achieved a low-risk rating.

    Of the 20 studies assessed using the JBI checklist [64] for analytical cross-sectional designs, 5 were rated as having low risk of bias [29,30,46,49,53], while the remaining 15 [26,31,32,36,38,42,50,51,54-59,68,70] were classified as moderate risk (Figure 4).

    To examine the distribution of fuzzy logic techniques across health care applications, a cross-tabulated synthesis was conducted. As shown in Figure 5, the most frequently applied approaches were FIS, FCM, and ANFIS, followed by FAHP, fsQCA, fuzzy evidential reasoning, and Takagi-Sugeno models. The use of these techniques varied across application domains, with oncology, infectious diseases, cardiovascular health, and mental health exhibiting the highest methodological diversity.

    The nature of causal engagement across the included studies spanned a continuum from explicitly formalized causal frameworks to approaches in which causal reasoning remained implicit or embedded within expert-driven or structurally defined fuzzy models. Only two studies (2/37, 5.4%) explicitly addressed causal questions using formal causal inference methodologies. A small subset relied on inferred causal structures derived from expert knowledge or fuzzy cognitive maps (6/37, 16.2%). In contrast, most studies primarily implemented predictive or associative modeling approaches, where causal interpretation was not formally specified and was instead inferred indirectly from model structure, expert judgment, or post hoc interpretation (29/37, 78.4%). This distribution highlights substantial heterogeneity in how causal principles are operationalized across fuzzy logic–based applications in health care.

    ‎
    Figure 4. Risk of bias across included studies assessed using PROBAST+AI and the JBI checklist. Among studies evaluated with PROBAST+AI, 53% (9/17) were classified as high risk and 47% (8/17) as moderate risk, with none rated as low risk. In contrast, among studies assessed using the JBI checklist, 25% (5/20) were rated as low risk and 75% (15/20) as moderate risk. JBI: Joanna Briggs Institute; PROBAST: Prediction Model Risk of Bias Assessment Tool.
    ‎
    Figure 5. Distribution of fuzzy logic techniques across health care domains. ANFIS: adaptive neuro-fuzzy inference system; FAHP: fuzzy analytic hierarchy process; FCM: fuzzy cognitive map; FER: fuzzy evidential reasoning; FIS: fuzzy inference system; fsQCA: fuzzy-set qualitative comparative analysis; FTT: fuzzy-trace theory; MCDM: multicriteria decision-making; TS: Takagi-Sugeno model.

    This systematic review synthesized evidence from 37 studies published between 2014 and 2025 that used fuzzy logic–based methodologies in health care settings with explicit or implicit causal objectives. The included studies span a wide range of clinical and public health domains, including infectious diseases, cancer, cardiovascular diseases, occupational health and safety, mental health, and preterm birth, underscoring the broad applicability of fuzzy modeling to diverse health-related problems. Across domains, the most frequently reported approaches were FIS, ANFIS, FAHP, and FCM. Rather than indicating methodological convergence, this distribution reflects context-dependent adaptations of fuzzy logic to address uncertainty, nonlinearity, and expert-guided reasoning in complex health care environments.

    Only a limited subset of studies conducted direct comparative evaluations between fuzzy logic–based models and conventional statistical or machine learning approaches. Among the 14 studies that included explicit comparators, 5 reported superior performance of fuzzy models [39,41,43-45]—most frequently in cancer and cardiovascular applications—while 3 demonstrated broadly comparable results [47,48,68]. The remaining 6 studies provided comparative analyses with insufficient methodological or statistical detail to support firm conclusions regarding relative effectiveness [28,40,57,59,66,67,69]. Importantly, most included studies relied on internal validation procedures, baseline comparisons, or expert-defined structures without external benchmarks, often using small- to medium-sized datasets. This pattern limits the generalizability of reported performance gains and indicates that, while fuzzy approaches may perform competitively in specific contexts characterized by nonlinearity or uncertainty, evidence supporting consistent superiority over conventional methods remains limited and heterogeneous.

    Causal inference was explicitly operationalized in only a small proportion of the included studies. Specifically, two investigations adopted formal causal inference frameworks: Lee et al [30] used fsQCA, explicitly modeling configurations of necessary and sufficient conditions at the population level. Mohandes et al [57] implemented a hybrid interval-valued intuitionistic fuzzy DEMATEL-ANP approach to structurally identify and prioritize causal drivers in occupational safety systems. In both cases, causal claims were grounded in transparent methodological procedures, explicit thresholds, and internally coherent validation strategies, rather than inferred post hoc from predictive performance.

    Beyond these two studies, causal reasoning was indirect. Six additional investigations relied on expert-based mappings, FCM, or influence structures to simulate causal mechanisms without formally testing necessity, sufficiency, or counterfactual dependence [28,38,44,53,54,66,69]. In most studies, fuzzy logic was applied primarily for predictive or associative purposes, with causal assumptions embedded implicitly within model architecture or domain expertise rather than explicitly articulated or empirically evaluated.

    Taken together, these findings reveal a marked disconnect between the theoretical capacity of fuzzy logic to represent causal structure and its prevailing empirical use in health care research. This gap appears to reflect not inherent conceptual limitations of fuzzy methods, but rather broader issues related to study design, validation practices, and reporting rigor, which constrain the translation of fuzzy modeling from predictive decision support to explicit causal inference.

    Risk of bias constituted a major limiting factor across the included studies. Among those evaluated with PROBAST+AI [65], none achieved a low-risk rating, with most classified as moderate or high risk, while only a small proportion of studies assessed using the JBI checklist [64] were rated as low risk.

    In contrast to much of the existing literature, which has primarily emphasized predictive accuracy or isolated clinical applications, the present review integrates formal risk-of-bias assessment with thematic synthesis to jointly evaluate reported performance, methodological rigor, and the explicitness of causal assumptions. This perspective highlights both the strengths and current limitations of fuzzy logic–based approaches: while they provide interpretable, rule-based models well suited to ambiguity and nonlinearity, their application within explicitly causal analytical frameworks remains limited and inconsistent.

    These conclusions must be interpreted considering several important limitations, including substantial heterogeneity across health care domains, modeling strategies, and outcome measures, which precluded quantitative meta-analysis; inconsistent reporting practices, such as limited use of comparator models and incomplete outcome reporting; and the frequent reliance on small- to medium-sized datasets without external validation. Collectively, these factors reduce the overall certainty and generalizability of the current evidence base.

    Despite these limitations, the findings carry important implications for both research and practice. Fuzzy systems appear particularly well suited to health care and policy contexts characterized by incomplete data, multidimensional interactions, and a strong demand for interpretability. Their capacity to encode expert knowledge and tolerate imprecision supports their use in applications such as risk stratification, early diagnosis, and context-sensitive prioritization. Realizing this potential, however, will require methodological consolidation, including greater standardization in reporting, more consistent use of comparator frameworks, and external validation across real-world datasets. Importantly, integration with formal causal frameworks—such as directed acyclic graphs or structural causal models—offers a pathway to strengthen causal interpretability while preserving the distinctive advantages of fuzzy reasoning.

    In parallel, recent advances in artificial intelligence have largely emphasized the automation of data extraction, measurement, and pattern recognition in clinical settings, particularly through machine learning and computer vision–based applications [71-76]. While these approaches have improved efficiency and scalability, they remain predominantly oriented toward prediction rather than causal inference. Addressing this gap requires analytical frameworks that move beyond automation to explicitly represent causal structure, intervention contrasts, and temporal assumptions.

    In this context, future research would benefit from explicitly incorporating TTE [17-20] when applying fuzzy logic to observational health care data. TTE provides a principled framework for specifying causal estimands, temporal ordering, and hypothetical interventions, thereby addressing key sources of bias that remain unresolved in many fuzzy-based applications. By defining eligibility criteria, treatment strategies, follow-up periods, and causal contrasts a priori, TTE can situate fuzzy, rule-based models within transparent causal designs—an approach that is particularly relevant in real-world health care settings where randomized trials are often infeasible.

    Viewed in this way, fuzzy logic should not be considered merely an auxiliary modeling technique, but a potential component of hybrid causal approaches in health care. When interpretability and causal structure are integrated into model design rather than treated as secondary considerations, fuzzy systems may help bridge the gap between statistical prediction and meaningful causal explanation. Advancing this agenda will require further methodological refinement, interdisciplinary collaboration, and a move toward more coherent and explicitly causal research programs in complex health systems.

    Acknowledgments

    The authors would like to express their gratitude to the Faculty of Medicine and the Faculty of Engineering of the University of Valparaíso for their institutional support, and to the Doctoral Program in Health Sciences and Engineering at the University of Valparaíso, Chile, for providing the academic framework in which this research was developed. The authors also wish to acknowledge Dr. Steren Chabert for her substantial intellectual contribution, guidance, and sustained support throughout the development of this research. In addition, the authors acknowledge the support of the National Agency for Research and Development (ANID), Government of Chile, through the Doctoral Scholarship Program (ANID – Subdirección de Capital Humano, Doctorado Nacional 2024, grant number 202421240031).

    Funding

    This work was supported by a doctoral scholarship from the National Agency for Research and Development (ANID), Government of Chile [Doctoral Scholarship Program, 2024, grant 1202/2024]. It resulted from a specific objective of the Sciences and Engineering for Health Program developed by the University of Valparaíso, Chile.

    Data Availability

    All data generated or analyzed during this study are included in this published article.

    Authors' Contributions

    JJ served as the principal author, leading the investigation and preparing the original draft of the manuscript. PM contributed to the investigation and conceptualization and critically revised the manuscript. KC-F participated in the investigation and conceptualization and contributed to the review and editing process. XC-T provided supervision and contributed to the development of the methodology. AV contributed to conceptualization, supervision, and methodological design.

    Conflicts of Interest

    None declared.

    Checklist 1

    PRISMA 2020 checklist.

    PDF File, 205 KB
    1. Seoni S, Jahmunah V, Salvi M, Barua PD, Molinari F, Acharya UR. Application of uncertainty quantification to artificial intelligence in healthcare: a review of last decade (2013-2023). Comput Biol Med. Oct 2023;165:107441. [CrossRef] [Medline]
    2. Alizadehsani R, Roshanzamir M, Hussain S, et al. Handling of uncertainty in medical data using machine learning and probability theory techniques: a review of 30 years (1991-2020). Ann Oper Res. Mar 21, 2021;339(3):1-42. [CrossRef] [Medline]
    3. Liu F. Data science methods for real-world evidence generation in real-world data. Annu Rev Biomed Data Sci. Aug 2024;7(1):201-224. [CrossRef] [Medline]
    4. Pearl J. Causality: Models, Reasoning and Inference. 2nd ed. Cambridge University Press; 2009. ISBN: 9780521895606
    5. Pearl J. An introduction to causal inference. Int J Biostat. Feb 26, 2010;6(2). [CrossRef] [Medline]
    6. Pearl J. On the consistency rule in causal inference: axiom, definition, assumption, or theorem? Epidemiology. Nov 2010;21(6):872-875. [CrossRef] [Medline]
    7. Pearl J. Graphical models, potential outcomes and causal inference: comment on Linquist and Sobel. Neuroimage. Oct 1, 2011;58(3):770-771. [CrossRef] [Medline]
    8. Rubin DB. Formal mode of statistical inference for causal effects. J Stat Plan Inference. Jul 1990;25(3):279-292. [CrossRef]
    9. Imbens GW, Rubin DB. Causal Inference for Statistics, Social, and Biomedical Sciences: An Introduction. Cambridge University Press; 2015. [CrossRef]
    10. VanderWeele TJ, Hernán MA. Causal inference under multiple versions of treatment. J Causal Inference. May 1, 2013;1(1):1-20. [CrossRef] [Medline]
    11. Moccia C, Moirano G, Popovic M, et al. Machine learning in causal inference for epidemiology. Eur J Epidemiol. Oct 2024;39(10):1097-1108. [CrossRef] [Medline]
    12. Gueltzow M, Bijlsma MJ, van Lenthe FJ. Beyond associations: from theory to interventions in health inequalities research using causal inference. Ann Epidemiol. Dec 2025;112:1-7. [CrossRef] [Medline]
    13. Sauré D, O’Ryan M, Torres JP, et al. Cost-savings and health impact of strategies for prevention of respiratory syncytial virus with nirsevimab in Chile based on the integrated analysis of 2019-2023 national databases: a retrospective study. J Infect Public Health. Apr 2025;18(4):102680. [CrossRef] [Medline]
    14. Hernán MA, Robins JM. Causal Inference: What If. Chapman & Hall Book; 2020. URL: https:/​/static1.​squarespace.com/​static/​675db8b0dd37046447128f5f/​t/​677676888e31cc50c2c33877/​1735816881944/​hernanrobins_WhatIf_2jan25.​pdf [Accessed 2026-03-03]
    15. Tennant PWG, Murray EJ, Arnold KF, et al. Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations. Int J Epidemiol. May 17, 2021;50(2):620-632. [CrossRef]
    16. Pearl J. Comments on: The tale wagged by the DAG. Int J Epidemiol. Jun 1, 2018;47(3):1002-1004. [CrossRef] [Medline]
    17. Hernán MA, Dahabreh IJ, Dickerman BA, Swanson SA. The target trial framework for causal inference from observational data: why and when is it helpful? Ann Intern Med. Mar 2025;178(3):402-407. [CrossRef] [Medline]
    18. Fu EL. Target trial emulation to improve causal inference from observational data: what, why, and how? J Am Soc Nephrol. Aug 1, 2023;34(8):1305-1314. [CrossRef] [Medline]
    19. Hernán MA, Robins JM. Using big data to emulate a target trial when a randomized trial is not available. Am J Epidemiol. Apr 15, 2016;183(8):758-764. [CrossRef] [Medline]
    20. Robinson CH, Fan E, Grandi SM, Urner M, Parekh RS. Target trial emulation in paediatric research: how can causal effects be estimated from observational data? Lancet Child Adolesc Health. Sep 2025;9(9):663-672. [CrossRef] [Medline]
    21. Cinelli C, Pearl J. Generalizing experimental results by leveraging knowledge of mechanisms. Eur J Epidemiol. Feb 2021;36(2):149-164. [CrossRef] [Medline]
    22. Zadeh LA. Fuzzy sets. Information and Control. Jun 1965;8(3):338-353. [CrossRef]
    23. Zadeh LA. Fuzzy systems theory: a framework for the analysis of humanistic systems. In: Cavallo R, editor. Systems Methodology in Social Science Research: Recent Developments. Springer Netherlands; 1982:25-41. [CrossRef]
    24. Zadeh LA. Fuzzy logic—a personal perspective. Fuzzy Sets and Systems. Dec 2015;281:4-20. [CrossRef]
    25. Mamdani EH, Assilian S. An experiment in linguistic synthesis with a fuzzy logic controller. Int J Man Mach Stud. Jan 1975;7(1):1-13. [CrossRef]
    26. Sarkar S, Singh P, Lingala MAL, Verma P, Dhiman RC. Malaria risk map for India based on climate, ecology and geographical modelling. Geospat Health. Nov 6, 2019;14(2). [CrossRef] [Medline]
    27. Sheela Rani M, Dhanasekar S. Application of type-2 heptagonal fuzzy sets with multiple operators in multi-criteria decision-making for identifying risk factors of Zika virus. BMC Infect Dis. Apr 1, 2025;25(1):450. [CrossRef] [Medline]
    28. Benito D, Quintero C, Aguilar J, Ramírez JM, Fernández-Anta A. Explainability analysis: an in-depth comparison between fuzzy cognitive maps and LAMDA. Appl Soft Comput. Oct 2024;164:111940. [CrossRef]
    29. Costa SB, Miranda CDS, De Souza BC, et al. Fuzzy and spatial analysis of cutaneous leishmaniasis in Pará State, Brazilian Amazon: an ecological and exploratory study. J Infect Dev Ctries. Jul 29, 2024;18(07):1124-1131. [CrossRef]
    30. Lee HY, Yang BM, Kang M. Factor configurations with governance as conditions for low HIV/AIDS prevalence in HIV/AIDS recipient countries: fuzzy-set analysis. J Korean Med Sci. Nov 2015;30:S167-S167. [CrossRef]
    31. Mollalo A, Khodabandehloo E. Zoonotic cutaneous leishmaniasis in northeastern Iran: a GIS-based spatio-temporal multi-criteria decision-making approach. Epidemiol Infect. Jul 2016;144(10):2217-2229. [CrossRef] [Medline]
    32. Piyatilake ITS, Perera SSN. Fuzzy multidimensional model to cluster dengue risk in Sri Lanka. Biomed Res Int. 2020;2020:2420948. [CrossRef] [Medline]
    33. Romero D, Olivero J, Real R, Guerrero JC. Applying fuzzy logic to assess the biogeographical risk of dengue in South America. Parasit Vectors. Sep 5, 2019;12(1):428. [CrossRef] [Medline]
    34. Shi X, Li J, Huang A, Song S, Yang Z. Assessing the outbreak risk of epidemics using fuzzy evidential reasoning. Risk Anal. Nov 2021;41(11):2046-2064. [CrossRef] [Medline]
    35. Boni TTA, Ayatollahi H, Langarizadeh M. A clinical decision support system for assessing the risk of cardiovascular diseases in diabetic hemodialysis patients. Curr Diabetes Rev. 2020;16(3):262-269. [CrossRef] [Medline]
    36. Aydın GZ, Özkan B. Evaluation of low-and middle-income countries according to cardiovascular disease risk factors by using Pythagorean fuzzy AHP and TOPSIS methods. BMC Med Inform Decis Mak. Nov 28, 2024;24(1):363. [CrossRef] [Medline]
    37. Korkmaz H, Canayaz E, Birtane Akar S, Altikardes ZA. Fuzzy logic based risk assessment system giving individualized advice for metabolic syndrome and fatal cardiovascular diseases. Technol Health Care. 2019;27(S1):59-66. [CrossRef] [Medline]
    38. Safaei M, A Sundararajan E, Asadi S. A hybrid MCDM approach based on fuzzy-logic and DEMATEL to evaluate adult obesity. Int J Environ Res Public Health. Nov 22, 2022;19(23):15432. [CrossRef] [Medline]
    39. Sabahi F. Bimodal fuzzy analytic hierarchy process (BFAHP) for coronary heart disease risk assessment. J Biomed Inform. Jul 2018;83:204-216. [CrossRef] [Medline]
    40. Yavari A, Rajabzadeh A, Abdali-Mohammadi F. Profile-based assessment of diseases affective factors using fuzzy association rule mining approach: a case study in heart diseases. J Biomed Inform. Apr 2021;116:103695. [CrossRef] [Medline]
    41. Saleh E, Błaszczyński J, Moreno A, et al. Learning ensemble classifiers for diabetic retinopathy assessment. Artif Intell Med. Apr 2018;85:50-63. [CrossRef] [Medline]
    42. Iancu I. Heart disease diagnosis based on mediative fuzzy logic. Artif Intell Med. Jul 2018;89:51-60. [CrossRef] [Medline]
    43. Subramanian J, Karmegam A, Papageorgiou E, Papandrianos N, Vasukie A. An integrated breast cancer risk assessment and management model based on fuzzy cognitive maps. Comput Methods Programs Biomed. Mar 2015;118(3):280-297. [CrossRef] [Medline]
    44. Mahmoodi SA, Mirzaie K, Mahmoodi MS, Mahmoudi SM. A medical decision support system to assess risk factors for gastric cancer based on fuzzy cognitive map. Comput Math Methods Med. 2020;2020:1016284. [CrossRef] [Medline]
    45. Yılmaz A, Arı S, Kocabıçak Ü. Risk analysis of lung cancer and effects of stress level on cancer risk through neuro-fuzzy model. Comput Methods Programs Biomed. Dec 2016;137:35-46. [CrossRef] [Medline]
    46. Brust-Renck PG, Reyna VF. Individual differences in numerical representations of risk in health decision making: a fuzzy-trace theory approach. Risk Anal. Mar 2023;43(3):548-557. [CrossRef] [Medline]
    47. Pota M, Scalco E, Sanguineti G, et al. Early prediction of radiotherapy-induced parotid shrinkage and toxicity based on CT radiomics and fuzzy classification. Artif Intell Med. Sep 2017;81:41-53. [CrossRef] [Medline]
    48. Stanković J, Stanković M. Estimation of the most influential factors for survival probability prediction of prostate cancer patients. Cancer Invest. Oct 21, 2017;35(9):594-600. [CrossRef] [Medline]
    49. Wolfe CR, Reyna VF, Widmer CL, et al. Efficacy of a web-based intelligent tutoring system for communicating genetic risk of breast cancer: a fuzzy-trace theory approach. Med Decis Making. Jan 2015;35(1):46-59. [CrossRef] [Medline]
    50. Barbounaki S, Sarantaki A. Construction and validation of a preterm birth risk assessment model using fuzzy analytic hierarchy process. Bosn J Basic Med Sci. Apr 1, 2022;22(2):291-299. [CrossRef] [Medline]
    51. Souza RT, Cecatti JG, Passini R Jr, et al. Cluster analysis identifying clinical phenotypes of preterm birth and related maternal and neonatal outcomes from the Brazilian multicentre study on preterm birth. Int J Gynaecol Obstet. Jul 2019;146(1):110-117. [CrossRef] [Medline]
    52. Salinas M, Velandia D, Mayeta-Revilla L, et al. An explainable fuzzy framework for assessing preeclampsia classification. Biomedicines. Jun 16, 2025;13(6):1483. [CrossRef] [Medline]
    53. Chen Y, Zhang L, Zhang J, et al. The path of depression in rural children: a fuzzy-set qualitative comparative analysis across multiple provinces in China. J Affect Disord. Nov 15, 2024;365:237-245. [CrossRef] [Medline]
    54. Hynek N, Franczukowska A, Rössl L, et al. A system model of post-migration risk factors affecting the mental health of unaccompanied minor refugees in Austria—a multi-step modeling process involving expert knowledge from science and practice. Int J Environ Res Public Health. Jul 14, 2020;17(14):5058. [CrossRef] [Medline]
    55. Upadhyay R, Kumar A, Bhattacherjee A, Patra AK, Chau N. Reducing musculoskeletal disorders in iron ore mine operators: a fuzzy-based intervention approach. WORK. 2024;78(1):131-152. [CrossRef] [Medline]
    56. Sakinala V, Paul PS, Fissha Y. Promoting safety of underground machinery operators through participatory ergonomics and fuzzy model analysis to foster sustainable mining practices. Sci Rep. Jul 15, 2024;14(1):16319. [CrossRef] [Medline]
    57. Mohandes SR, Karasan A, Erdoğan M, Ghasemi Poor Sabet P, Mahdiyar A, Zayed T. A comprehensive analysis of the causal factors in repair, maintenance, alteration, and addition works: a novel hybrid fuzzy-based approach. Expert Syst Appl. Dec 2022;208:118112. [CrossRef]
    58. Malakoutikhah M, Zare A, Karimi A, Hassanipour S. Fuzzy logic modeling of factors affecting musculoskeletal disorders in a steel factory in Iran: a cross-sectional study. J Manipulative Physiol Ther. Mar 2021;44(3):221-228. [CrossRef] [Medline]
    59. Demir E, Ceyhun Sabır E. Evaluation of physical risk factors by fuzzy failure mode and effects analysis: an apparel mill example. Int J Occup Saf Ergon. Mar 2025;31(1):89-98. [CrossRef] [Medline]
    60. Li S, Chu Z, editors. Machine Learning for Causal Inference. Springer; 2023. [CrossRef]
    61. Liu T, Krentz A, Lu L, Curcin V. Machine learning based prediction models for cardiovascular disease risk using electronic health records data: systematic review and meta-analysis. Eur Heart J Digit Health. Jan 2025;6(1):7-22. [CrossRef] [Medline]
    62. Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al. Cochrane Handbook for Systematic Reviews of Interventions. Cochrane; 2024. URL: https://training.cochrane.org/handbook [Accessed 2026-03-03]
    63. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. Mar 29, 2021;372:n71. [CrossRef] [Medline]
    64. Barker TH, Habibi N, Aromataris E, et al. The revised JBI critical appraisal tool for the assessment of risk of bias for quasi-experimental studies. JBI Evid Synth. Mar 1, 2024;22(3):378-388. [CrossRef] [Medline]
    65. Moons KGM, Damen JAA, Kaul T, et al. PROBAST+AI: an updated quality, risk of bias, and applicability assessment tool for prediction models using regression or artificial intelligence methods. BMJ. Mar 24, 2025;388:e082505. [CrossRef] [Medline]
    66. Amirkhani A, Mosavi MR, Mohammadizadeh F, Shokouhi SB. Classification of intraductal breast lesions based on the fuzzy cognitive map. Arab J Sci Eng. May 2014;39(5):3723-3732. [CrossRef]
    67. Maranate T, Pongpullponsak A, Ruttanaumpawan P. The prioritization of clinical risk factors of obstructive sleep apnea severity using fuzzy analytic hierarchy process. Comput Math Methods Med. 2015;2015:257856. [CrossRef] [Medline]
    68. Argyropoulos A, Townley S, Upton PM, Dickinson S, Pollard AS. Identifying on admission patients likely to develop acute kidney injury in hospital. BMC Nephrol. Feb 14, 2019;20(1):56. [CrossRef] [Medline]
    69. Sümbül-Şekerci B, Pasin Ö, Egeli D, Gönenç S, Şekerci A. Characterizing cognitive phenotypes and clinical correlates in type 2 diabetes using fuzzy clustering and decision tree analysis. Sci Rep. Oct 14, 2024;14(1):23965. [CrossRef] [Medline]
    70. Scrobota I, Iova GM, Marcu OA, et al. An artificial intelligence-based fuzzy logic system for periodontitis risk assessment in patients with type 2 diabetes mellitus. Bioengineering (Basel). Feb 20, 2025;12(3):211. [CrossRef] [Medline]
    71. Antel R, Whitelaw S, Gore G, Ingelmo P. Moving towards the use of artificial intelligence in pain management. Eur J Pain. Mar 2025;29(3):e4748. [CrossRef] [Medline]
    72. Calderone A, Latella D, Fauci EL, et al. Mind meets machine: a narrative review of artificial intelligence role in clinical psychology practice. Clin Psychol Psychother. 2025;32(6):e70191. [CrossRef] [Medline]
    73. Chen K, Bai Y, Xie X. Dynasmile: AI-powered smile analysis software. Int Dent J. Oct 2025;75:105483. [CrossRef]
    74. Luo X, Li Y, Xu J, Zheng Z, Ying F, Huang G. AI in medical questionnaires: scoping review. J Med Internet Res. Jun 23, 2025;27:e72398. [CrossRef] [Medline]
    75. Aravazhi PS, Gunasekaran P, Benjamin NZY, et al. The integration of artificial intelligence into clinical medicine: trends, challenges, and future directions. Dis Mon. Jun 2025;71(6):101882. [CrossRef] [Medline]
    76. Fathima M, Moulana M. Revolutionizing breast cancer care: AI-enhanced diagnosis and patient history. Comput Methods Biomech Biomed Engin. Apr 2025;28(5):642-654. [CrossRef] [Medline]

    ‎
    ANFIS: adaptive neuro-fuzzy inference system
    AUC: area under the curve
    DEMATEL-ANP: decision-making trial and evaluation laboratory analytic network process
    FAHP: fuzzy analytic hierarchy process
    FCM: fuzzy cognitive map
    FIS: fuzzy inference system
    fsQCA: fuzzy-set qualitative comparative analysis
    GRADE: Grading of Recommendations Assessment, Development, and Evaluation
    JBI: Joanna Briggs Institute
    MeSH: Medical Subject Headings
    PICO: population, intervention, comparator, and outcome
    PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses
    RQ: research question
    TTE: target trial emulation

    Edited by Fabian Prasser; submitted 03.Sep.2025; peer-reviewed by Anup Gupta, Ke Chen, Shashank Mirji; final revised version received 23.Jan.2026; accepted 12.Feb.2026; published 25.Mar.2026.

    Copyright

    © Jaime Jamett, Patricio Melendez, Ximena Collao-Ferrada, Karina Cordero-Torres, Alejandro Veloz. Originally published in JMIR AI (https://ai.jmir.org), 25.Mar.2026.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR AI, is properly cited. The complete bibliographic information, a link to the original publication on https://www.ai.jmir.org/, as well as this copyright and license information must be included.