Transforming care for the elderly with AI-powered insights

Benete, a Finnish company and IBM partner specializing in elder care technology, sought to revolutionize how caregivers and nurses leverage Assisted Daily Living (ADL) insights to deliver top-tier care to elderly patients. A significant challenge in elder care services is the lack of digitalization, overwhelming caregivers with administrative tasks. This also complicates the detection of early changes in functional ability or cognitive decline, especially in older individuals who live alone. Even with frequent nurse visits, subtle changes in daily activities may be missed, making it difficult to offer preventive care.

Benete’s BeneCare solution addressed these challenges by leveraging IoT sensor technology and intelligent algorithms to remotely and unobtrusively monitor elderly clients’ well-being to provide a holistic view of their daily lives. BeneCare would then generate diverse data points around subtle changes in their activity levels, physical movement, sleep patterns, hygiene, food intake and preparation, and other routines. These data points also helped caregivers detect and prevent falls. BeneCare's dashboards provided caretakers and nurses with rich, comprehensive data, which they manually analyzed and derived insights from. However, the time-strapped nature of their work hindered thorough analysis and translation of the information into actionable reports and summaries. The manual processes resulted in underutilization of valuable data, risking oversight of critical health changes and causing negative impacts on the quality of care. Benete, recognizing the need for a solution that automates the creation of summaries to optimize BeneCare’s usability and maximize their contribution to better patient outcomes, set out to develop this capability.

Benete and IBM® Client Engineering embarked on a four-week pilot to enhance the functionality of BeneCare dashboards—by leveraging IBM watsonx.ai®. With access to large language models (LLMs), specifically Llama and Mistral, Jupyter Notebook and the watsonx.ai Python software development kit, Benete saved valuable time in documentation procedures, enabling caregivers to focus more on patients.

By creating prompt templates in watsonx.ai, Benete sourced variables, data and tasks from IBM Cloudant®, a scalable JavaScript Object Notation (JSON) database. This database not only stored insights from IoT sensors and delivered structured data findings and summaries but also queried the data and summaries efficiently.

Benete tracked the development process and monitored various LLM metrics within the prompt templates generated with Jupyter Notebooks. These prompt templates were then tracked using AI factsheets created by IBM watsonx.governance® to monitor and evaluate the different AI models. Additionally, these AI factsheet-tracked prompt templates were saved separately so that Benete could log the number of tokens used by the care unit and track their costs—a valuable feature in an embedded solution. These prompts were then deployed in watsonx.ai to become endpoints where data is transformed into understandable insights and integrated into BeneCare.

By integrating all these components, Benete successfully created and automated user-friendly summaries from its data using watsonx.ai, which can also be translated into various languages despite time and language constraints. The BeneCare dashboard continuously monitors key care metrics and generates data 24x7. Summary reports, which were previously generated and interpreted manually, are now instantly generated in less than 5 seconds using this data, empowering care professionals to make informed decisions and provide the best care possible. Llama and Mistral were the LLMs chosen to be deployed on watsonx.ai due to their size, cost benefits and positive evaluations of the way reports were written to best address Benete’s needs. Having tested the efficacy of both models, Benete is able to keep the functionality model-agnostic, allowing for flexibility in LLM development. 

The IT transformation has yielded substantial benefits for Benete. The LLM-based summary add-on in BeneCare helps caregivers understand their elderly clients’ condition and capabilities easily, prompting tailored, preventive care and attention, while the alerts enable timely intervention. This helps elderly individuals remain in their homes for as long as they wish, as care providers can offer preventive care while also managing a growing number of clients. The solution also helps Benete with report localization, generating translatable and tone-sensitive content that facilitates seamless communication across multicultural care environments. With such easily translatable reports, caregivers from diverse backgrounds can share a unified understanding of client needs, improving care quality.

With the transformed solution in place, Benete’s day-to-day operations run smoother and are more streamlined. Care teams are freed from manual data extraction and interpretation thanks to rapid access to summaries, saving time and boosting efficiency. Notably, there’s a 90% reduction in the time taken to understand client data, as nurses now take just one minute—not ten, like before— to understand charts. This reduction translates to 30 minutes of daily time savings per nurse, as this exercise is performed three times a day per client.

What Benete needed was not a chatbot to address queries, but the automation of a complex, time-consuming task, freeing up users to focus on their core competencies—elderly care and support. With the transformed BeneCare platform, caregivers can use LLMs to accelerate qualitative data interpretation and craft summaries that can be readily customized as needed, ensuring that information is conveyed in the most decision-ready format.

In a traditional, labor-forward industry, Benete is a front-runner, utilizing a critical solution to interpret data and derive insights for an expedited reporting process. Now, after the successful pilot, Benete can provide a platform capable of being customized based on the unique needs of end-users. Benete’s ultimate goals are to further enhance patient outcomes, provide better preventive care and ease administrative burdens on care staff.

Benete (link resides outside of ibm.com) is a Finland-based software company specializing in technology for efficient senior care, providing timely and accurate data to organizations and individuals for personalized care. Its services include cost-saving preventive care models, enabling high care efficiency and improving the quality of life of the elderly.