Sign in Subscribe
Artificial intelligence

Artificial Intelligence in Healthcare: Evaluating the Feasibility of Personalized Rehabilitation Programs

Strategic Insights

— 1 min read

Researchers from Peninsula Health in Australia have published a new study evaluating the use of artificial intelligence (AI) and large language models (LLMs) in generating tailored rehabilitation programs for patients undergoing major head and neck surgical procedures. The study assessed the feasibility of using LLMs to create personalized rehabilitation plans, comparing the quality, accuracy, and clinical relevance of plans generated by four different AI models. The findings indicate that AI-generated rehabilitation plans may complement existing protocols, but further clinical validation is necessary to assess their impact on patient outcomes.

Key Takeaways:

  • The study evaluated the use of four AI models (ChatGPT-4o, DeepSeek V3, Gemini 2, and Copilot) in generating tailored rehabilitation programs for patients undergoing major head and neck surgical procedures.
  • The AI models were prompted with identical queries to generate rehabilitation plans, which were then assessed by three senior clinicians using a five-point Likert scale.
  • ChatGPT-4o achieved the highest clinical relevance (Likert mean of 4.90 ± 0.32), followed by DeepSeek V3 (4.00 ± 0.82) and Gemini 2 (3.90 ± 0.74), while Copilot underperformed (2.70 ± 0.82).
  • Gemini 2 produced the most readable content, according to the study's readability and quality metrics.
  • The study highlighted the potential for AI-generated rehabilitation plans to complement existing protocols, but emphasized the need for further clinical validation to assess their impact on patient outcomes.

Statistics:

  • 10 hypothetical head and neck surgical clinical scenarios were developed to test the AI models.
  • Four AI models (ChatGPT-4o, DeepSeek V3, Gemini 2, and Copilot) were evaluated in the study.
  • The clinical relevance of the rehabilitation plans was assessed using a five-point Likert scale, with ChatGPT-4o achieving a mean score of 4.90 ± 0.32.
  • Gemini 2 produced the most readable content, according to the study's readability metrics, with a Flesch Reading Ease score of 70.9 ± 2.1.

Sources:

  • Leveraging Artificial Intelligence for Personalized Rehabilitation Programs for Head and Neck Surgery Patients. Technologies, 2025,13(4):142. (Technologies - http://www.mdpi.com/journal/technologies)
  • MDPI AG (publisher)
  • NewsRx. Research on Artificial Intelligence Described by Researchers at Peninsula Health (Leveraging Artificial Intelligence for Personalized Rehabilitation Programs for Head and Neck Surgery Patients). Medical Devices & Surgical Technology Week. May 18, 2025; p 972.

Read more