Related Experiment Video
Updated: Nov 25, 2025
06:37
Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
Published on: December 15, 2023
4.8K
Identifying mental health status using deep neural network trained by visual metrics
Somayeh B Shafiei1,2, Zaeem Lone1,2, Ahmed S Elsayed1,2
1Applied Technology Laboratory for Advanced Surgery (ATLAS), Roswell Park Comprehensive Cancer Center, Buffalo, NY, USA.
Translational Psychiatry
|December 15, 2020
Summary
This study introduces an objective AI model using CNN-LSTM algorithms to monitor cancer patients
Area of Science:
- Artificial Intelligence
- Machine Learning
- Computer Vision
Background:
- Mental health is crucial for cancer patients' quality of life.
- Depression and anxiety are prevalent in cancer patients, potentially leading to severe outcomes like suicide.
- Current objective methods for mental health evaluation are lacking, relying mostly on subjective patient-therapist interactions.
Purpose of the Study:
- To introduce an objective method for evaluating mental health in cancer patients.
- To develop a model capable of categorizing mental health metrics (hope, anxiety, mental well-being) using AI.
- To enable home-based mental health monitoring for early identification of at-risk patients.
Main Methods:
- A combination of Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) algorithms was employed.
- Visual metrics time-series data were recorded using TobiiPro eyeglasses from cancer patients and a control group.
- Pre- and post-study questionnaires (HHI, STAI, WEMWBS) were administered to assess mental health metrics.
Main Results:
- The proposed CNN-LSTM model achieved high classification accuracy: 93.81% for Herth Hope Index (HHI), 94.76% for State-Trait Anxiety Inventory for Adults (STAI), and 95.00% for Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS).
- The model successfully objectified and categorized the mental health status of individuals based on the assessed metrics.
- The system demonstrated the potential for reliable, objective mental health assessment.
Conclusions:
- The developed AI model offers an objective approach to mental health evaluation for cancer patients.
- This technology can be integrated into home-based monitoring applications for post-oncologic surgery patients.
- Early identification of at-risk patients can be facilitated, enabling timely interventions.
