Patient classification using machine learning
Diagnosis by measuring skin conductance
Diagnosis of mental illness through wireless communication

A domestic research team has succeeded in developing a technology that objectively diagnoses the condition and severity of depression patients using a skin conductance sensor. This opens the way to diagnosing and early predicting the onset of depression using biosignals.
The Electronics and Telecommunications Research Institute (ETRI) announced on the 29th that the results of an experiment conducted on depression patients in collaboration with the team of Professor Jeon Hong-jin of the Department of Psychiatry at Samsung Seoul Hospital using a skin conductance sensor capable of measuring minute sweat changes on the skin, noting that depression patients' sweat response becomes dull when they are stressed, were published in Scientific Reports.

The research team said that since the diagnosis and treatment of patients with mental illnesses such as depression have mainly relied on psychological tests or medical questionnaires, they began the study to provide medical professionals with a more objective method for early detection and prevention of mental illnesses.
When the mental state of patients worsens, hormonal responses related to the brain become disrupted, which leads to autonomic nervous system responses. The research team began research and development to objectively measure physiological changes, such as minute amounts of sweat, using a skin conductance sensor attached to the fingertips to help doctors make more efficient diagnoses.

Afterwards, a noninvasive (diagnosing or treating a disease without penetrating the skin or passing through a hole in the body) biosignal data measurement was performed on people without depressive disorder, patients with major depressive disorder, and patients with panic disorder for three months. In this paper, it was explained that it is possible to distinguish between patients with depressive disorder and those without depressive disorder, targeting approximately 60 mental patients with major depressive disorder at the Department of Psychiatry at Samsung Seoul Hospital, including those without depressive disorder.

In particular, the research team revealed in this paper that it is possible to diagnose depressive disorder using skin conductance signals, and further developed an automatic diagnosis model based on machine learning that can objectively and more accurately diagnose and monitor the condition of depressive disorder patients.
The research team emphasized that in order to predict disease signs more accurately, it is necessary to supplement the analysis techniques based on complex sensors such as brain waves, heartbeat, respiration, and temperature in addition to skin conductance. As a result, if the level of research completion is improved in the future, it is expected that it will be possible to diagnose and predict symptoms of various mental illnesses such as panic disorder, attention deficit hyperactivity disorder (ADHD), trauma, and autism in addition to depression.

To this end, ETRI conducted additional data analysis on four elements other than the skin conductance sensor through joint research with the team of Professor Jeon Hong-jin of the Department of Psychiatry at Samsung Seoul Hospital. Through follow-up observation, the patient’s diagnosis and psychological test results, blood and sweat, heart rate, respiration, blood pressure, and brain wave bio-signal data were obtained.
In addition, the research team created a composite module (sensor) capable of measuring multiple biosignals measuring 36.5 mm x 33 mm. The research team said that although the sensor is ready for immediate commercialization, the sensor size needs to be reduced and the level of completion needs to be increased to make it into a wearable device with wireless communication in order to reach a level where it can be applied to actual patients in the future. The plan is to apply it to the wrist through a wearable device in the future.

Therefore, if the results of this study are commercialized, it is expected that sensors can be attached to wearable watches to analyze sweat and measure blood pressure and heart rate. This will allow people to identify their condition at an early stage. If applied to patients in the future, it is expected that it will be possible to automatically notify guardians or hospitals of serious conditions and manage them.
This study was developed over a period of three years from 2015 through the “Development of a skin-attachable sensor module for monitoring and predicting symptoms of mental illness” project of the Ministry of Science and ICT. Through this technology development, the research team applied for three international patents and 18 domestic patents. The number of SCI papers published reached 17.

ETRI Biomedical IT Research Center Director Seung-Hwan Kim, who is in charge of the research, said, “We have seen the potential for developing a biosignal-based mental illness diagnosis and prediction system that can objectively diagnose and predict mental illness.”

The research team plans to develop a technology that can predict early signs as well as objectively diagnose patients with mental illness by applying biosignal data to machine learning in the future. They also plan to produce additional research results on the classification of mental illness based on blood and complex biosignals.

Meanwhile, the main author of this paper is researcher Kim A-young of the Biomedical IT Research Center of ETRI. The research team of Professor Jeon Hong-jin of the Department of Psychiatry at Samsung Seoul Hospital and Professor Byeon Sang-won of the Department of Electronic Engineering at Incheon National University also participated in the research.