Japanese IT company Fujitsu has developed a digital twin solution, using machine learning and generative AI, to simulate the social impact of local government healthcare policies. The product, called Policy Twin, identifies measures to reduce costs and improve outcomes in preventive healthcare, the company said. In field tests, it was able to “identify policy candidates” that doubled both cost savings and health improvements, it said. It anticipates that, “across multiple municipalities”, it can generate best practices and lead to the standardization of policies.
A statement explained: “[The solution] recreates previously successful local government policies on a digital twin and uses data to generate new policy candidates and gauge effectiveness. During field trials… [it] enabled the creation of policy candidates that doubled both the medical expense cost-savings and the improvement of health indicators achieved through health guidance compared to the previous year, while meeting resourcing requirements. This demonstrated [its] ability to create policies that achieve multiple objectives simultaneously, potentially shortening planning times and facilitating consensus building.”
Fujitsu has been developing a group of technologies under the banner Social Digital Twin to incorporate behavioral economics into digital services. It has a mission to support “solutions for complex societal challenges”, which hinges on a “digital rehearsal technology” to recreate human behaviour on a digital twin and predict the effect/impact of policies. The new Policy Twin builds on this concept, it said, by using “empirical economics”, described as the use of data science to best-allocate resources to “preemptively assess and compare the effectiveness of digitally recreated local government policies on a digital twin”.
The process, as explained, is more prosaic, as always: : to convert publicly available municipal policy documents into a machine-readable flowchart format; to generate new flowchart candidates by crossing and comparing successful flowcharts from multiple municipalities; and to simulate service provision for each newly-generated flowchart candidate. All of it is assisted by large language models (LLMs) and machine learning (ML). The service is available for testing by municipal agencies in Japan, starting next week (December 6). Fujitsu aims to launch the service for the healthcare and medical sector in Japan “by fiscal year 2025”, it said. It is unclear how or if it will be cascaded into international markets.
A statement said: “Municipalities [will be able] to achieve improvements in resident health, cost savings, and disease prevention, ultimately contributing to improved wellbeing. As the Policy Twin can also demonstrate the rationale behind the proposed policies, it can facilitate consensus-building among diverse stakeholders and support its implementation in society. Furthermore, employing this technology across multiple municipalities is expected to lead to best practice creation, mutual referencing of policies between municipalities, and application towards the standardization of policies.”
