A Computer-Based Platform for Aiding Clinicians in Eating Disorder Analysis and Diagnosis
May 10th, 2022
Diagnosing eating disorders in healthcare is challenging. Therefore, the present protocol develops an algorithm based on 949 patient responses to a questionnaire, with the diagnosis displayed on an easy-to-use web-based interface. This system facilitates the accurate diagnosis of eating disorders while excluding those believed to have an eating disorder.
The system decides if a patient fulfills the diagnostic criteria for an eating disorder and what type of eating disorder such as anorexia nervosa, bulimia nervosa or binge eating disorder. So patients tend to be more honest when answering a computer than a physical person. And, of course, the system is saving time for the clinician.
The purpose of the method is to improve the reliability of eating disorder diagnosis, assisting clinicians in selecting those who suffer from an eating disorder and referring those who do not fulfill eating disorder criteria to other treatments or no treatments. Demonstrating the procedures will be Alkioni Glibi, a research assistant at our laboratory. Upon patient referral, navigate to the web landing page using any modern browser.
Use an existing account associated with a clinician to log into the web tool. Fill in the patient registration form, including patient ID, birth date, age, and sex, social security number. Press the Save button to register a new patient.
Open the questionnaire application on a smart device. Fill in the social security number and first visit date for the patient. The current date is used by default.
Then fill in the information corresponding to the patient's weight, height and age. Next, fill in the information corresponding to behavior such as induced vomiting, snack frequency and eating rate, followed by particulars corresponding to cognitive and emotional items, such as fear of gaining weight and feelings of body dysmorphia. Press the Done button to finish the questionnaire.
Navigate to the web landing page using any web browser. Use an existing account to log into the web tool. Search for the patient using the patient's social security number or patient ID.Add measured weight and height to the system.
Press the Result tab to get the algorithmic decision of whether the patient has an eating disorder, or ED and what type of ED.Then press the tab questions one through 20 or questions 21 through 34 to display the questions where the patient responses deviate from answers by healthy individuals. Select a final diagnosis under the Result tab, based on the algorithm and the clinician's expertise. An example of a risk assessment page for recommended diagnosis and the estimated probability of accuracy from zero to one is shown here.
From the recommended diagnosis and questionnaire responses, healthy and deviating responses were generated. The algorithm estimating the probability of different types of eating disorders for the individual is demonstrated here. The accuracy of the model was determined to be 97.1 for ED and 82.8 for ED diagnosis.
Another benefit of the system is that it provides the answers responsible for the suggested diagnosis, which allows the system to teach clinicians to diagnose patients better and to consult additional health professionals in difficult cases.
Overview
This article presents a web-based algorithm designed to improve the diagnosis of eating disorders by analyzing patient responses to a questionnaire. The system enhances diagnostic accuracy while allowing clinicians to efficiently assess patients.
Key Study Components
Area of Science
- Clinical psychology
- Healthcare technology
- Eating disorder diagnosis
Background
- Diagnosing eating disorders is often challenging for healthcare providers.
- Traditional methods may lead to misdiagnosis or underdiagnosis.
- Utilizing technology can improve patient honesty and streamline the diagnostic process.
- The algorithm is based on extensive patient data to enhance reliability.
Purpose of Study
- To develop a reliable diagnostic tool for eating disorders.
- To assist clinicians in identifying patients who meet the criteria for eating disorders.
- To provide a user-friendly interface for clinicians and patients.
Methods Used
- Web-based questionnaire for patient responses.
- Algorithmic analysis of responses to determine diagnosis.
- Integration of patient data such as weight, height, and behavioral patterns.
- Clinician input for final diagnosis based on algorithm results.
Main Results
- The algorithm achieved a diagnostic accuracy of 97.1% for eating disorders.
- It provided insights into patient responses that deviated from healthy norms.
- Clinicians received recommendations based on algorithmic assessments.
- The system aids in teaching clinicians better diagnostic practices.
Conclusions
- The web-based tool significantly improves the reliability of eating disorder diagnoses.
- It allows for better patient management and referral processes.
- The system can enhance clinician training and decision-making.
Frequently Asked Questions
Executive Industry Relevance
Algorithm-driven diagnostic platforms for complex behavioral disorders address a critical gap in early, accurate identification, especially in primary care settings lacking specialist expertise. By standardizing diagnostic criteria and quantifying probability estimates, such systems enhance predictive confidence and streamline patient triage. This approach supports scalable, reproducible decision-making across diverse clinical environments, directly impacting portfolio-level risk management in digital health and behavioral medicine R&D.
Strategic Applications in Biopharma R&D
Early Discovery & Target Validation
- Enables systematic interrogation of diagnostic criteria for behavioral disorders using large-scale, expert-annotated datasets.
- Supports biological and behavioral de-risking by quantifying diagnostic probability and highlighting response deviations.
- Facilitates predictive confidence in patient stratification and early intervention strategies.
Screening & Assay Development
- Prepares validated digital phenotyping systems for downstream clinical and translational workflows.
- Standardizes data capture and algorithmic assessment for reproducible, quantitative outputs.
- Enables scalable screening of at-risk populations with consistent diagnostic thresholds.
Translational & Preclinical Research
- Aligns digital diagnostic outputs with clinical endpoints for translational continuity.
- Supports risk-adjusted advancement of digital biomarkers and behavioral endpoints.
- Provides mechanistic de-risking by linking questionnaire responses to diagnostic outcomes.
Pipeline & Workflow Integration
This computer-based diagnostic platform integrates into the early discovery-to-clinical translation continuum, supporting hypothesis testing, digital biomarker validation, and scalable patient stratification.
- Discovery Biology: Quantifies diagnostic probability and identifies key response deviations for hypothesis-driven refinement.
- Screening: Delivers reproducible, algorithm-based outputs for high-throughput patient assessment.
- Analytics: Provides quantitative readouts and statistical accuracy metrics to inform cross-condition comparisons.
- Translational Research: Bridges digital diagnostic data with clinical decision-making and referral pathways.
- Enterprise Reuse: Offers a reusable, web-based infrastructure adaptable to additional behavioral or psychiatric indications.
Operational & Enterprise Impact
- Scientific Value: Increases predictive confidence and reduces diagnostic ambiguity in behavioral health.
- Operational Value: Standardizes and automates diagnostic workflows for scalability and reproducibility.
- Strategic Value: Enables earlier, data-driven go/no-go decisions and optimizes resource allocation.
- Portfolio Impact: Supports risk-adjusted prioritization of digital health assets and diagnostic algorithms.
Implementation Considerations
- Requires clinical expertise for final diagnosis and interpretation of algorithmic outputs.
- Needs secure digital infrastructure for patient data capture and analysis.
- Demands cross-team standardization of data entry and interpretation protocols.
- Adaptable to various clinical settings but may require localization for different populations.
- Dependent on the quality and representativeness of training datasets for algorithmic accuracy.
