At Mind Moves, we are committed to using Artificial Intelligence (AI) for the greater good and driving positive societal impact. We actively engage in initiatives that promote AI education and responsible development to create a more inclusive and equitable AI future.
But what does that mean exactly? Well, we just wrapped up our latest Generative AI (GenAI) pilot program with one of our government clients, the National Library for Medicine (NLM) – one of the 27 institutes at the National Institutes of Health (NIH). It is a shining example of our process and the results we produce shared here to help you as you consider your AI innovation journey.
Pilot Goal
Our objectives coming into this pilot were to test large language models (LLMs) for the benefit of internal process improvement, while also developing workforce skills and fostering a supportive community of practice.
The Process
We took a structured research approach to dive into this new frontier. First, we asked a diverse staff of 18 innovators – from non-technical fellows to expert coders and PhDs – to clearly describe their goals and what they hoped to achieve with GenAI. Next, we helped them define the methods they would use and the metrics to measure their progress and success.
Of course, as we innovate, we must prioritize safety and ethical considerations to ensure AI advancements benefit everyone responsibly. To that end, we do not use open web versions of LLMs. Instead, we operate within secure, firewalled platforms like Microsoft Azure’s AI playground.
Project Objectives
The 10 pilot projects spanned several categories; including, responsive customer service, operational process efficiencies, summarization and analysis of large data sets, and clinical support assistance tools. Here are a few highlights of the projects that sought to improve accessibility, biomedical discovery and health outcomes using GenAI.
- To boost website accessibility to comply with 508 standards and expedite the resolution of common errors, we employed an automatic detection and resolution program.
- For consumer-health related inquiries, we used LLMs grounded in data to provide a first draft response to customer questions. This has the potential to improve efficiency in customer response handling and customer satisfaction levels.
- One team streamlined metadata mapping and dataset ingestion with Python-coded scripts to reduce costs and give us the ability to scale repositories in the collection to accelerate biomedical data discovery.
- We deployed semantic-based models to investigate the relationship between data sequences and literature and their impact on health outcomes to drive biomedical computational advancements.
- Additionally, we architected a robust method demonstrating high prediction accuracy and ranking of criteria for inclusion in clinical trials using patients’ health record notes, promising a more efficient and accurate method for matching patients to clinical trials.
Impact of Generative AI Pilot
Organizational Change
This pilot signified a culture of innovation and aligned well to NLM’s strategic vision and initiatives. The 18 pilot innovators representing every division took advantage of technical support and resources provided, they participated in about 50 cohort meetings and more than 300 one-on-one consultations to build their skills and their drive prototype development. The cohort members also had training sessions, joined Hackathons, and benefited from the momentum and synergy of learning in a cohort experience. Along side every step of development, safety measures guided by the NIST AI Risk Management Framework were considered.
This work has generated a great deal of attention both internally and externally. We had a broad-reaching impact across NLM and NIH to help other staff trying to dip their toes into the AI/LLM waters. The pilot was also featured in several publications, including FedScoop, underscoring the significant impact this work has had in just a few short months.
Impact of Innovation
Through their dedication and scientific approach, the innovators were able to prove that GenAI can be successfully applied to both solve their hypothesis and enhance their productivity. Research and development type innovation in AI has its challenges, but the rewards are worth the pursuit. Everyone involved learned new skills, supported each other’s progress, and created a palpable energy that propelled the entire organization forward in its AI journey.
Moreso, as these projects advance, the potential for their impact on health outcomes is immeasurable. We are increasing access to trusted information, improving response time to customers, and matching patients to new medical treatments. In turn, we can reach traditionally underserved communities and improve the quality of care and life for hundreds of thousands, maybe even millions of people around the globe.
We Can Help
Mind Moves takes pride in our ability to skillfully navigate the AI landscape and help clients to structure a path to achieve their mission. Building a successful pilot program in today’s era of AI is way beyond the traditional “paint by numbers” approach. Emergent technologies and methods are constantly evolving. Our talented team leads from the cutting-edge of developments, stretching leaders and employees to rethink how to solve their problems using safe, human-driven AI solutions. Please contact us if we can be of service!
