CareLink: An AI-Driven Integrated Healthcare-HR Platform for Coordinated Management of Learning Disorders and Workplace Support

Abstract

CareLink is a privacy-aware healthcare-HR platform for coordinated workplace support of employees with specific learning disorders. The prototype screens dyslexia and dyscalculia using employee-consented questionnaire inputs, functional-impact indicators, persistence/history items, and consistency guardrails; dysgraphia is not included in this pilot. Two disorder-specific XGBoost classifiers generate dyslexia and dyscalculia risk probabilities. Confidence and quality-control rules combine these outputs into low-risk, disorder-specific risk, combined-risk, or low-confidence escalation categories. High-risk and low-confidence cases are reviewed by clinicians before any accommodation recommendation is shared with HR. A controlled pilot with 87 participants used internal cross-validation, clinician-confirmed labels, workflow timestamps, questionnaires, and interviews. The dyslexia and dyscalculia models achieved AUC values of 0.93 and 0.90. Mean time-to-support decreased from 34.7 to 8.3 days, and usability was high (SUS = 80.4). The contribution is a clinician-governed decision-support workflow that connects AI screening with privacy-preserving HR accommodation coordination without replacing diagnosis or automating employment decisions.