National Implementation of an Artificial Intelligence-Based Virtual Dietitian for Patients With Cancer

Area of Science:

• Oncology
• Nutrition Science
• Health Informatics

Background:

• Nutritional status significantly impacts cancer patient outcomes.
• A shortage of registered dietitians limits access to essential nutritional counseling for cancer patients.
• Artificial intelligence (AI) and machine learning (ML) offer potential solutions to enhance nutritional support accessibility.

Purpose of the Study:

• To develop and evaluate an AI-based nutrition assistant to provide guideline-based nutritional support to cancer patients.
• To assess the feasibility and patient-reported benefits of implementing an AI virtual dietitian.

Main Methods:

• Developed an AI nutrition assistant, Ina, using a dataset of over 100,000 expert-curated interventions, literature, and clinical guidelines.
• Ina provides a conversational text message interface for dietary guidance and answering patient questions.
• Implemented Ina nationally in partnership with 25 advocacy organizations, collecting data on demographics, outcomes, and utilization.

Main Results:

• 3,310 users registered nationally between July 2019 and August 2023, with diverse cancer types and high medical complexity.
• Users reported prevalent nutritional challenges, including overweight/obesity and barriers to good nutrition.
• High engagement and satisfaction were observed, with 94% satisfied and 98% finding guidance helpful.
• Significant benefits reported: 82% improved quality of life, 88% improved symptom management, and 84% used advice to guide diet.

Conclusions:

• An evidence-based AI virtual dietitian (Ina) is feasible to implement.
• Patients reported benefits in diet, quality of life, and symptom management.
• Ongoing evaluations are underway to assess the impact on additional outcomes.