Korean researchers have developed artificial intelligence (AI) capable of detecting Parkinson’s disease in its early stages, when the condition shows no clear symptoms and diagnosis is typically difficult.

Samsung Seoul Hospital’s AI Research Center announced that it has confirmed the potential for early diagnosis and prognosis prediction of neurodegenerative diseases, including Parkinson’s disease and Parkinson-plus syndromes. The team utilised multimodal AI technology that analyses brain imaging tests such as MRI and PET scans alongside various clinical data including gait and voice patterns.

Parkinson’s disease is a neurodegenerative disorder in which dopamine-producing neurons in the brain’s substantia nigra are destroyed, causing movement impairment. It is considered one of the three major geriatric brain diseases along with dementia and stroke.

Early symptoms are often unclear, frequently delaying diagnosis. By the time patients experience trembling hands and feet, muscle stiffness, slowed movement, and gait abnormalities, the disease has likely already progressed significantly.

Even specialists find it difficult to distinguish early-stage ‘Parkinson-plus syndromes’ – conditions such as progressive supranuclear palsy and multiple system atrophy that present similar symptoms but have different causes.

AI capable of detecting subtle patterns

The research team, led by Professor Cho Jin-hwan of the Neurology Department and Professor Chung Myung-jin of the Radiology Department at Samsung Seoul Hospital, found a solution in AI capable of detecting subtle pattern differences invisible to the human eye.

Over four years, they collected gait, voice, and brain imaging data from approximately 500 patients, including 363 with Parkinson’s disease, 67 with progressive supranuclear palsy, and 61 with multiple system atrophy. The data was standardised to build an integrated database.

Based on this foundation, the team developed a gait data-based fall risk prediction model, a voice test-based Parkinson’s classification system, and an MRI-based automated brain structure analysis model.

Clinical evaluation showed the voice-based severity classification model achieved an accuracy (AUC) of 0.96, while the MRI-based disease differentiation model recorded 0.91. The fall prediction model combining gait and brain imaging analysis also demonstrated strong performance with an accuracy of 0.84.

The newly developed AI does not simply produce results – it also presents the basis for its judgments. The researchers enabled automatic selection of gait stability indicators, brain structural changes, and voice characteristics to serve as diagnostic evidence.

Additionally, because the system was built on a dedicated data storage and analysis system (NAS) within the hospital’s internal network, AI analysis is possible without external transfer of medical data. This secures both privacy protection and research efficiency.

Earlier detection, more effective treatment

“The earlier Parkinson’s disease is detected, the more effective drug treatment becomes, and rehabilitation can slow its progression,” said Professor Cho Jin-hwan. “AI will help with early diagnosis by rapidly synthesising multiple test results and contribute to developing personalised treatment plans for each patient.”

Professor Chung Myung-jin said, “Based on this research, we will expand applications to other neurological diseases such as dementia and develop multi-institutional collaborative studies to help more patients.”

Through this research, Samsung Seoul Hospital’s AI Research Center has published 27 SCIE-level papers and filed 45 patents. The developed technologies are being utilised for follow-up research across more than 10 clinical departments, including emergency medicine, ophthalmology, and rehabilitation medicine.

“We plan to continue building an integrated AI research platform, developing disease-specific AI models, and pursuing open innovation with global companies and research institutions,” said Yang Kwang-mo, Director of the AI Research Center. “We will lead the practical application of AI across various fields.”

Source:

Original article by Ahn Kyung-jin, Medical Correspondent, Seoul Economic Daily