A Machine Learning Approach to Detect Depression in an Individual

According to a population-based study in India, the prevalence of depression stands at 15.1%, affecting an estimated 57 million individuals in the country. This accounts for roughly 18% of the global population. However, a significant obstacle to timely detection of depression cases is the severe scarcity of psychiatrists, amounting to a staggering 77%. To address this issue, a self-assessment method has been developed, allowing individuals to assess whether they might be experiencing depression without the need for a psychiatrist's visit. This method can also be employed by psychiatrists to augment their diagnostic process and ensure prompt patient care. The self-assessment process involves a user interface that presents a series of questions. The responses are captured through speech acoustics, eye blink rate, and EEG signals. These data are then pre-processed and analysed individually using trained machine learning (ML) and convolutional neural network (CNN) models. The final outcome determines if an individual is experiencing depression. Additionally, the system generates a comprehensive report that includes acoustic spectrogram features, components, individual analyses, and outputs.