Artificial intelligence (AI) technology is being applied across a wide range of life science research. In particular, it plays a critical role in improving the diagnostic accuracy of degenerative neurological diseases and refining treatment strategies. It is expected that personalized treatment might also be possible by simulating drugs found by AI using stem cells derived from patient cells.
On the 7th, at the ‘Golden Triangle Life Science Open Innovation’ forum held at COEX in Seoul, scientists from South Korea, the United Kingdom, and Japan presented the latest research achievements utilizing bio and digital technologies. This forum was organized as a special session of Bio Korea 2025.
First, Haruhisa Inoue, a principal researcher at the iPS Cell Research Institute (CiRA) at Kyoto University in Japan, presented a case of using induced pluripotent stem cells (iPS cells) and AI in diagnosing and treating amyotrophic lateral sclerosis (ALS). ALS, also known as Lou Gehrig’s disease, is a rare condition where all muscles in the body gradually become paralyzed, and no fundamental treatment has been developed yet.
Principal researcher Inoue introduced an AI named ‘PM-HDE’ that finds drug candidates which are effective in treatment using iPS cells from ALS patients. iPS cells are cells that have been reprogrammed to an embryonic stem cell state, capable of growing into any cell type in the human body by inserting specific genes, proteins, or chemical substances into fully grown cells.
The AI PM-HDE predicts the structures of compounds suitable for therapeutic agents based on thermal diffusion equations and ranks potential drug candidates. The research team tested about 1,000 compounds designed by AI on nerve cells created from ALS patients' iPS cells. As a result, they selected 29 candidate drug substances. These compounds are undergoing development as drugs in pharmaceutical companies after passing cellular experiments.
A deep learning algorithm for early diagnosis of ALS has also been developed. Deep learning is a method of deep machine learning that mimics the neural structure of humans. The research team trained AI with data from motor neurons obtained from ALS patients, enabling the analysis of nerve cell differentiation images later to predict the likelihood of ALS onset. Inoue said, ‘It can be used as an auxiliary tool for ALS diagnosis,’ and noted, ‘It can also contribute to developing future treatment strategies.’
Principal researcher Inoue also identified blood-based biomarkers capable of classifying sporadic ALS based on AI trained on gene expression data. Sporadic ALS is caused by environmental factors, accounting for about 90% of patients. He added, ‘What has been discussed so far is research related to ALS, but in the future, it can also be applied to develop treatments for various diseases by applying new datasets.’
Kay Cho, a professor of neuroscience at King’s College London in the UK, presented research results on diagnosing degenerative neurological diseases like Alzheimer’s disease and elucidating their onset processes using AI. Professor Cho’s research lab has been utilizing AI machine learning since 2013. Machine learning is an AI technology that learns large volumes of data and autonomously finds methods without prior programming.
As a result of AI learning from data on degenerative neurological diseases, they found that the synapses of patients had significantly weakened. Synapses are the consolidation points that transmit signals from one nerve cell to another. They also revealed that tau proteins play an important role in synaptic weakening.
Based on this, the research team designed peptide drugs that bind to tau proteins using AI. Peptides are fragments of proteins. Professor Cho explained, ‘AI proposed peptide structures that can stably move to the target and be expressed,’ adding, ‘It can save time in designing suitable peptides.’
Additionally, they confirmed that degenerative changes in nerve cells occur gradually by training AI on 10,000 images related to degenerative neurological diseases. Professor Cho stated, ‘AI is an excellent tool, but biological validation is essential,’ and mentioned, ‘We developed a method to validate AI’s predictive results using brain cortical tissue.’ He further commented, ‘If South Korea, Japan, and the UK collaborate as one team, we can achieve better results.’
Yoon Hyung-jin, a professor in the Department of Biomedical Engineering at Seoul National University, discussed patient-centered treatment in the era of generative AI. Professor Yoon said, ‘Patient-centered treatment encompasses three elements: communication, partnership, and health promotion, with quantum communication being especially important,’ explaining, ‘Just as doctors and patients decide on treatment options through discussion, the correct information must be well shared between doctors and patients.’
To achieve this, digital twin technology can be utilized. A digital twin is a virtual representation designed to accurately reflect real-world objects. Simulations using this technology can predict what results might occur when a specific treatment is performed, and based on the predictions, discussions with patients can be held to determine treatment methods.
Professor Yoon emphasized, ‘In this process, it is important to enhance health literacy so that patients can understand the meanings of simulation results,’ noting, ‘Comparing generative AI like ChatGPT with human doctors shows that ChatGPT scored high in terms of information quality and empathy, indicating that large language models like ChatGPT can change the landscape of healthcare.’
Large language models, a type of AI machine learning method for understanding human language, can also play a major role in advancing digital twin technology. Professor Yoon remarked, ‘Collecting data from patients for use in the digital twin can yield patient-tailored data, and although there is still a long way to go, I believe it can be achieved through collaboration.’