Evaluation of academic stress among medical students using graphology and machine learning algorithm in correlation with salivary cortisol

Abstract

Background: Stress is a part of the academic life of graduates. Young adults are especially susceptible to academic stress based on their subjective commitment towards academic goals, and social pressure for superior academic performance. Recognizing academic stress is crucial for planning successful management, and to prevent mental illness. The aim of the study was to develop methods to identify stress using graphology, machine learning algorithm and salivary cortisol.Methods: The study included a mixed method research design and enrolled 43 medical students (19 males and 24 females) between 18-23 years of age were taken in the present study. The Kessler psychological distress scale (K10) was used to ascertain distress among the study subjects.Results: Students written manuscript images were taken for artificial intelligence training and analysis. The written manuscript was evaluated for positive and negative personality traits using graphology techniques. One of the negative traits identified by graphology, i.e.; dejection, it was used to train a machine learning tool to identify the negative trait of dejection.Conclusions: This study suggests that mental health professionals can train machine learning algorithms, using graphology tools, to function as a screening tool to determine stress levels. This may help to plan for management and recovery of stressed individuals and help them in future academic performance.