Symptom mapping and personalized care for depression, anxiety and stress: A data-driven AI approach

Abstract

Background

Depression, anxiety, and stress disorders have significant and widespread impacts worldwide, affecting millions of individuals and their communities. According to the World Health Organization, depression impacts the daily lives of more than 300 million people, making it one of the most important diseases globally. Treatment for these mental disorders (MD) typically involves medication and psychotherapies, but also incorporates technological resources like Artificial Intelligence (AI) to indicate personalized therapies and care. While various AI approaches have been applied in the context of MD in the literature, they often focus solely on aiding diagnosis.

Objective

This research proposes an AI approach for mapping symptoms and assisting in the personalized care of depression, anxiety, and stress.

Methods

Symptom mapping utilizes data mining (DM) techniques to generate rules representing knowledge extracted from data of 242 patients collected using the Depression, Anxiety, and Stress Scale (DASS-21). This knowledge elucidates how symptoms impact the severity degrees of considered MDs. Subsequently, the generated rules are employed to construct a Fuzzy Inference System (FIS) for inferring the severities of MDs based on patient symptoms and personal data.

Results and conclusions

The results achieved in the DM (accuracy ≥92.98 %, sensibility ≥86.02 %, specificity ≥97.32 %, and kappa statistic ≥87.98 %), indicating consistent patterns, along with the results produced by the FIS, demonstrate the potential of the proposed approach to assist health professionals in rapidly predicting symptoms of depression, anxiety, and stress, thereby facilitating outpatient screening and emergency care. Furthermore, it can improve the association of symptoms, referral to specialized care, therapeutic proposals, and even investigations of other diseases unrelated to MD.