Depression is one of the leading causes of disability worldwide. Even though depression is a treatable condition, it's still underdiagnosed and under-treated. Due to a strong social stigma associated with mental illness, patients are reluctant to seek professional help, which often leads patients to suffer in silence and untreated. In addition, a misconception about mental health and superstitious belief in the Malaysian community contribute to the causes.
What it does
Therefore, instead of relying on self-reporting of depression, a subconscious way of assessment is highly needed. Studies showed that handwriting/graphology can reveal one's psychological and physiological, hence widely used as a valuation tool in medical and psychological diagnosis. Therefore, this research is designed to propose an automated graphology analysis using machine learning techniques for identifying signs of depression. Our work hence focuses on the graphological signs that can be mapped to the clinical signs of depression
How we built it
To the best of our knowledge, there is no such study for computational methods in graphology analysis for assessing signs of depression in Malaysia particularly. Hence, we propose this research initiative for identifying early signs of depression using the automated graphology analysis.
Challenges we ran into
Getting the data related to this research is very challenging.
Accomplishments that we're proud of
Phase 1-Design a codebook from the graphology signs to automate the feature extraction. Phase 2- Develop an automated model based on the topic modeling technique to classify the graphology signs.
What we learned
1) To analyze the graphology features that are significant to the signs of depression. 2) To develop a codebook of depression-graphology by incorporating the graphological signs and clinical symptoms of depression. 3) To construct a feature extraction model in depression-graphology using machine learning technique.
What's next for A Graphology-based Depression Model for Early Diagnosis
This project output could be useful as a new automated diagnostic tool for assessing signs of depression towards an early diagnosis of depression. The automated tool could provide a great complement to mental health care in traditional settings which aims to improve the identification of mental health problems. Hence, the prevalence of underdiagnosed and under-treatment of depression could be reduced significantly.