Case Study: Empowering Mental Health Awareness with AI-Driven Voice Sampling

Empowering Mental Health Awareness with AI-Driven Voice Sampling

In a bustling college campus in Karnataka, students juggle academic pressures, social expectations, and personal challenges daily. Mental health often takes a backseat—ignored, undiagnosed, or simply misunderstood. Enter Manodayam, a pioneering AI-driven mental health assessment initiative that sought to change this narrative through an innovative approach: Manodayam Artificial Intelligence Model.

A leading Engineering College, Shivamogga, established in 1980, is a premier engineering institution with 4,000+ students. Known for academic excellence and holistic development, it partnered with Manodayam’s AI-driven mental health initiative to address student well-being. With state-of-the-art infrastructure and industry ties, A leading Engineering College emphasizes innovation and mental health, shaping a brighter future for its students.

The Challenge: A Silent Crisis

Mental health challenges in Indian universities are significant, with studies indicating that nearly 50% of students experience stress, anxiety, or depression. A 2022 survey by the Indian Journal of Psychiatry revealed that 65% of university students reported academic pressure as a major stressor, while 30% faced financial stress. Despite this, only 10-15% seek help due to stigma and lack of awareness. The National Mental Health Survey (2015-16) highlighted that 1 in 20 Indians suffers from depression, with youth being highly vulnerable. Universities often lack adequate counseling services, with a student-to-counselor ratio of 1:100,000, far below global standards, underscoring the urgent need for systemic interventions.

Depression is often overlooked until it reaches severe stages. Many students hesitate to seek help due to stigma, lack of awareness, or fear of judgment. Traditional self-reporting methods, like the PHQ-9 questionnaire, help, but they have limitations—they rely on honest self-assessment, which may not always reflect a person’s true emotional state.

Manodayam believed voice holds untapped potential in mental health screening. Changes in tone, pitch, and speech patterns can indicate stress, anxiety, and even depression. However, the AI needed training—and for that, Manodayam needed real data from real students.

The Initiative: Building a Safe Space for Mental Health Analysis

Understanding the sensitivity of mental health discussions, Manodayam first gained the trust and approval of the college administration. With ethical considerations at the forefront, the process was structured to ensure privacy and voluntary participation.

Step 1: Educating & Engaging Students

An interactive seminar introduced students to the concept of AI-based mental health screening. Experts explained how their voices, when analyzed scientifically, could help create a non-invasive, stigma-free approach to mental health detection. Participation was voluntary, and students were assured of confidentiality.

Step 2: Manodayam Artificial Intelligence Model

Each participating student first completed the Manodayam AI Model Questions, a clinically validated tool for screening depression. Immediately after, they provided voice samples by reading out scripted sentences and answering a few open-ended emotional prompts.

The AI was trained to analyze:

  • Tone & Pitch: Variations that may indicate emotional distress
  • Speech Pauses & Pace: Slow speech could indicate low energy, often linked with depression

Vocal Strain: Higher strain levels can indicate stress or anxiety

Data Insights: Unveiling Hidden Trends

Key Observations & Impact

  • 61% of students showed some signs of depression (Mild to Severe levels).
  • 13% of students fell in the ‘Moderate to Severe’ category, requiring immediate attention.
  • Males had a higher count in Severe Depression (13 out of 15 cases), emphasizing the need to encourage men to open up about mental health.
  • Borderline depression was the most common non-normal category, affecting 95 students (25.6%).
  • A small but significant portion of students (9) preferred not to specify gender, highlighting the importance of inclusive mental health strategies.

Outcomes: More Than Just Data

  • Enhanced AI Model Accuracy: The collected dataset helped train the AI to better recognize vocal biomarkers associated with depression.
  • Raised Awareness: Students became more conscious of their mental well-being and were encouraged to seek professional help if needed.
  • Encouraged Conversations: The study initiated campus-wide discussions on the importance of mental health.
  • Potential for Early Interventions: The data allows educators and mental health professionals to develop targeted wellness programs.