TOKYO -- Crunching a vast amount of medical data for signs of illness takes time and effort, but recent technological advances in artificial intelligence are set to make the process more efficient.
Japanese conglomerate Hitachi and Tokyo's Keio University are among those at the forefront of this research. Unlike conventional data analysis, AI can detect signs of multiple diseases, including heretofore unknown warning signs.
Hitachi has started operating a pilot program to predict the risk of multiple lifestyle-related diseases on a group of employees, finding that AI can play a major role in lowering medical costs.
Hitachi's system analyzes about 150 pieces of medical data, seeking signs of 20 diseases, including diabetes, high blood pressure and hardening of arteries. It predicts a person's five-year chance of developing these illnesses, allowing companies to encourage lifestyle changes among employees.
Out of about 210,000 group employees, Hitachi studied the results of 45,000 employees with poor checkup results. The AI system was able to narrow that down to 50 employees who could benefit from guidance on their health. After six months of coaching, the employees' medical expenses declined by about 13,000 yen ($107) on average from the previous six months. But the savings only reached a little under 2,000 yen for 600 employees randomly chosen from the pool of 45,000.
Hitachi plans to further verify the efficacy of the system and will consider selling it to outside companies.
Meanwhile, at Keio University, a team led by Professor Tadahiro Kuroda has succeeded in developing an AI capable of detecting lung cancer in urine with about 90% accuracy.
Focusing on the fact that the composition of urine changes when a person develops lung cancer, the team has developed an AI that identifies about 400 compounds and their quantity, fishing out lung cancer-specific features.
Often, X-ray exams fail to detect lung cancer. Keio's AI has an added advantage of not requiring special facilities. The team is considering commercializing the system through partnership with the private sector.
According to the National Federation of Health Insurance Societies, annual premiums grew by about 100,000 yen per person between fiscal 2009 and 2015. The Ministry of Health, Labor and Welfare has been instructing health insurance societies across Japan to screen medical data to encourage lifestyle changes.