🚨 Pulse of Innovation 🚨

Breaking news in the healthcare AI

Claude 3 is a new winner…wait, don’t get sucked in marketing gimmicks

• Social media is flooded with messages that we have a new winner in LLM (large language model). “Claude 3 outperforms GPT4”, “New LLM destroyed GPT4” etc. While we are excited about the growth in the field of AI, let us also be realistic. As clinicians, we decided to see if this is really meaningful for our clinical practice. I use ChatGPT daily; should I now sign up for another service?

• Realistically, the answer is NO! There might be advantages in particular use cases, like generating codes with fewer hallucinations than GPT4. If you go by statistical methodology, the confidence interval completely overlaps the overall model strength, arguing that the difference is not better in all sectors. Now, if you really want to try it out, Claude 3 gives you some basic credits, so go check it out! With no COI, I am going to stick to ChatGPT use for now 🙂 

🧑🏼‍🔬 Bench to Bedside👨🏽‍🔬

Developments in healthcare AI research and innovations

AI-Human collaborative approach to improve surgical consent readability

• Surgical consent forms are often complex and cumbersome to follow for an average patient and may be time-consuming. AI-human collaborative approach may be beneficial.

• Consent forms from multiple institutions were assessed for readability and simplified using GPT-4. Medical authors and malpractice defense attorneys conducted independent reviews of the simplified consent. The GPT-4 was then used to generate subspecialty-specific de novo surgical consent, and it was again evaluated based on an 8-point metric and a subspecialty expert.

• Analysis of 15 academic medical centers’ consent forms revealed significant reductions in average reading time, word rarity, and passive sentence frequency following GPT-4-facilitated simplification. Readability improved from an average college freshman to an 8th-grade level, matching the average American’s reading level.

  

   

👩‍⚕️AI in the Clinic 🏥

Real-world and practical use of healthcare AI in clinic

How AI reduces first-year nursing staff turnovers in half?

• Laudio, a web-based nursing management platform, uses AI and automation to provide nursing leaders with valuable insights into their frontline team members' attendance, scheduling, staffing, and other information. The nursing team leaders can now access all this relevant information on a single platform rather than spending time searching for it on multiple platforms.

• Nebraska Health, an academic medical institution, implemented the Laudio platform and observed a decrease in the time their nursing team leaders spent on administrative tasks, improved efficiency, increased team leaders' timely interactions with team members, and a 50% reduction in first-year nurse turnover rates.

  

🩺 Start-Up Stethoscope 💵 

Trending startups and technologies impacting clinical practice

Generative AI and “Note Taking”: A rival to Microsoft powered Dax Copilot?

• Abridge AI has generative AI software that automates Clinical Note taking based on clinical conversation, a technology very similar to Dax Copilot highlighted in last edition of AI Grand Rounds. Sutter Health a prominent California based health system recently announced implementation of Abridge AI for clinical note taking for their providers.

• Abridge AI has secured consistent funding over the past two years, with the most recent funding of $150 million by Lightspeed Venture Partners, Redpoint Ventures, and CVS Health Ventures in February. Implementing Sutter Health could be a huge boost for their technology, which rivals Dax.

  

🤖Patients First, AI Second🤖

Ethical and regulatory landscape of healthcare AI

Whitehouse announces policy and oversight for AI use for all federal agencies including CDC and VA

• Although this policy announcement applies to all federal agencies, it includes key federal healthcare agencies like the CDC and VA, which will be required to appoint a Chief AI Officer and establish various safeguards for AI use and implementation.

• These safeguards include a wide array of "mandatory actions to reliably assess, test, and monitor AI’s impacts on the public, mitigate the risks of algorithmic discrimination, and provide the public with transparency into how the government uses AI." It also seeks to improve access to AI technology by bolstering workforce and reducing unnecessary barriers.

• The new fact sheet, for example, cites AI's potential to advance public health, noting that the Centers for Disease Control and Prevention is using AI to predict the spread of disease and detect the illicit use of opioids, while the Centers for Medicare and Medicaid Services is using the technology to reduce waste and identify anomalies in drug costs.