Sleep-EVAL Knowledge-Based Clinical Evaluation

Current architecture updated | September 2026
Summary by Maurice M. Ohayon, MD, DSc, PhD

Sleep-EVAL is the sleep-disorders KBS. Diagnostic reasoning is performed by the single integrated Ad-Infer system, which uses Sleep-EVAL knowledge during the clinical interview and inferential process.

FROM QUESTIONNAIRE TO CLINICAL INFERENCE

A conventional computerized questionnaire follows predetermined questions or branches. A knowledge-based clinical system must do more: it must decide which information is still needed, formulate and test clinical hypotheses, evaluate alternative explanations, and determine whether the available evidence legitimately supports a diagnosis.

Sleep-EVAL supplies Ad-Infer with the specialized sleep knowledge required for this process. The knowledge base encodes diagnostic definitions, criteria, exclusions, timing, severity and frequency constructs, related medical and psychiatric conditions, medications and other contextual elements relevant to sleep-disorder evaluation.

AD-INFER CLINICAL REASONING

Ad-Infer is a hybrid artificial intelligence system designed for structured clinical interviewing, phenotyping and diagnostic reasoning. Within Ad-Infer, EVAL-KBS combines domain-specific knowledge with Type-2 causal reasoning, fuzzy evaluation, Hopfield-based recurrent neural computation, learning processes and Bayesian belief updating. Symptoms and clinical states can therefore be represented as graded, interacting constructs rather than being reduced to simple present/absent variables.

The inferential process extends beyond recognition of positive criteria. Ad-Infer evaluates direct and indirect causal relationships, alternative explanations, comorbid conditions, medication and treatment effects, and required exclusion criteria. It can ask additional questions when evidence is insufficient, reopen an earlier hypothesis when contradictory information appears, and strengthen a positive diagnosis when new evidence is concordant.

INTEGRATED LANGUAGE UNDERSTANDING

Language-model capabilities are integrated within Ad-Infer. They are not a separate diagnostic agent. Ad-Infer uses them to extend its understanding of the clinical interaction: whether questions are asked or skipped, understood or misunderstood, accepted or refused, and how clarifications or unexpected responses relate to the current causal model. Returned information is interpreted again through the EVAL-KBS reasoning and belief model.

This bidirectional process allows the interview itself to generate clinically useful evidence. Confirmatory evidence may reinforce the positive diagnosis; ambiguous or contradictory evidence may reduce confidence, modify the causal interpretation, or become part of the differential-diagnosis process.

WHY THIS MATTERS FOR EPIDEMIOLOGY

The same structured clinical logic can be applied across large population samples, multiple sites, languages and repeated assessments. This reduces omissions, improves comparability and preserves clinically interpretable reasoning while generating detailed phenotypes for epidemiological and longitudinal analyses.

SELECTED REFERENCES

  1. Ohayon M. Validation of expert systems: Examples and considerations. Medinfo. 1995;8:1071-1075.
  2. Ohayon MM. Improving decision making processes with the fuzzy logic approach in the epidemiology of sleep disorders. J Psychosom Res. 1999;47:297-311.
  3. Ohayon MM, Guilleminault C, Zulley J, Palombini L, Raab H. Validation of the Sleep-EVAL system against clinical assessments of sleep disorders and polysomnographic data. Sleep. 1999;22:925-930.
  4. Ohayon M. Knowledge Based System Sleep-EVAL: Decisional Trees and Questionnaires. Bibliothèque Nationale du Québec / Bibliothèque Nationale du Canada. ISBN 2-921483-06-8; 1995.