Nhu Quynh started researching parkinson's disease in mid-2021 when she was a third-year student with the desire to have a new method to support doctors in diagnosing the disease. She learns about AI application models to detect parkinsonism through EEG. Doctors' current diagnosis methods are mainly based on the patient's clinical symptoms such as shaking limbs, difficulty talking, or falling.

Quynh said, symptoms of parkinson's disease in the early stages can be confused with other diseases. The AI ​​model will help doctors diagnose early and accurately, allowing patients to receive more timely treatment.

According to her, the main cause of parkinson's disease is that the brain loses an internal neurotransmitter, hindering communication between neurons. This can be determined through changes in brain waves, so this is the basis for her to build a disease diagnosis model.

Because she did not have the conditions to get EEG data from patients, Quynh used a data set of EEG parameters of 56 people from a foreign university and hospital. Among them, there were 28 people with parkinson's disease, in two stages with medication treatment before electroencephalography and without medication. The remaining 28 people were not sick. From this data, Quynh classifies, cleans and feeds into the machine learning model. Test reading results are more than 96% accurate.

The author said that the most important step is data preprocessing because the brain's electrical signals are quite sensitive and easily affected by external noise signals. Currently, Quynh's parkinson's disease prediction model is in the stage of using available data, and has not yet built a database from actual patients. Building a process for collecting EEG data requires specialized equipment and procedures from the hospital, so this is a limitation that the study has not been able to implement.

In addition to predicting parkinson's disease, Quynh believes that electrical brain signals can serve as a basis for diagnosing a number of diseases such as epilepsy, convulsions, neurological disorders... If support is received from doctors. When the hospital provides the patient's electrical brain data, the team can build a more detailed and accurate prediction model.

In addition to predicting parkinson's disease, Quynh believes that electrical brain signals can serve as a basis for diagnosing a number of diseases such as epilepsy, convulsions, neurological disorders... If support is received from doctors. When the hospital provides the patient's electrical brain data, the team can build a more detailed and accurate prediction model of the severity of parkinson's disease.

"Currently the model only gives results of having or not having the disease, so a lot of data is needed for the model to become intelligent and correctly predict what stage the patient is in," Quynh said.

According to Associate Professor, Dr. Huynh Van Tuan, Head of the Department of Physics and Engineering Physics, University of  Sciences, Ho Chi Minh City National University, using AI models to diagnose Parkinson's disease is a potential direction for major results. accurate based on the database. He said that in the near future, the school will cooperate with hospitals to share data sources for this research. 

Associate Professor, Dr. Nguyen Ngoc Lam, former Director of the Institute of Electronics - Informatics - Automation Research (Ministry of Industry and Trade), said that the model needs to be implemented on thousands or tens of thousands of patients to have a basis for evaluation. The scientific price is more valuable.

The main article is published on the electronic newspaper Vnexpress. Readers can see the link below:

https://vnexpress.net/sinh-vien-lam-mo-hinh-chan-doan-benh-parkinson-bang-dien-nao-do-4575853.html