Page 91 - AI for Good Innovate for Impact
P. 91

AI for Good Innovate for Impact



               Use Case Status: Ongoing, actively deployed and undergoing continuous
               development�

               Partners: Ministry of Health and Welfare [5], National Rehabilitation Centre, Gimhae City [6],
               UNIMEWO [14], Eisai Korea [10], SpiralAI (Japan) [7], Yonsei University [8], Seoul National          4.1-Healthcare
               University [13], Dongguk University [9], Konkuk University [11]


               2�2     Benefits of use case

               •    Enhance cognitive and physical health through accessible AI content
               •    Reduce emotional stress and supports social interactions


               2�3     Future Work

               •    ESTsoft is committed to advancing its AI-based senior care solutions through a multi-
                    faceted roadmap that combines technological innovation with domain expertise. Central
                    to this effort is the development of a senior-specific LLM that incorporates linguistic,
                    cultural, and contextual patterns unique to elderly users, enabling more intuitive and
                    empathetic interactions. In parallel, the company is enhancing its STT and TTS technologies
                    to more accurately reflect emotional nuance and accommodate the vocal characteristics
                    of older adults. These advancements are supported by an expanding network of Proof-
                    of-Concept (PoC) validations conducted both domestically and internationally, ensuring
                    service effectiveness across diverse user environments. To enable broader adoption, the
                    AI services are being delivered as scalable SDKs, compatible with multiple devices.
               •    ESTsoft is also reaching out to accelerate its global expansion through strategic
                    partnerships, beginning with active collaborations in Japan and the United States.
                    Underpinning these technological developments is a rigorous content development
                    process carried out in close collaboration with leading institutions: physical exercise
                    modules are co-developed with Yonsei University [13], safety education with the Korea
                    Safety Education Association [12], and cognitive programs with Eisai Korea [11] and
                    Boramae Medical Center at Seoul National University [13]. The validity and impact of
                    this expert-driven approach have been substantiated through academic research
                    and publications in peer-reviewed journals, reinforcing the credibility and real-world
                    applicability of ESTsoft’s AI-powered care solutions.


               3      Use Case Requirements

               •    REQ-01: It is critical that a dedicated LLM is trained on senior caregiving data to ensure
                    context-aware, empathetic, and relevant conversational responses for elderly users.
               •    REQ-02: It is critical that the platform provide a UI/UX specifically optimized for elderly
                    users, focusing on simplicity, clarity, and accessibility to accommodate cognitive and
                    motor limitations.
               •    REQ-03: It is critical that the system utilises highly accurate STT technology adapted to
                    aged voices, accounting for the slower speech, pronunciation variations, and volume
                    differences common among seniors.
               •    REQ-04: It is critical that a  personalised content recommendation engine be implemented
                    to tailor educational and engagement materials based on each user’s health status and
                    interaction history.
               •    REQ-05: It is critical that the service be compatible across multiple devices, including PCs,
                    TVs, tablets, and mobile phones, via a scalable SDK, ensuring accessibility regardless of
                    hardware.
               •    REQ-06: It is expected that a structured tagging system based on health classifications will
                    be applied to all content, allowing for accurate and consistent matching with user needs.





                                                                                                     55
   86   87   88   89   90   91   92   93   94   95   96