Guest: Dr. Stephon Proctor from Children's Hospital of Philadelphia Hosts: Adam Carewe and Dale Gold

Overall Summary: In this episode, Dr. Stephon Proctor discusses his journey from clinical psychology to informatics, the implementation of AI in healthcare, and the future of technology in the medical field. He shares insights on the challenges and successes of integrating AI tools to improve efficiency and reduce clinical burden.

Section Summaries

Introduction [00:00:08]

The hosts introduce the podcast and the guest, Dr. Stephon Proctor, highlighting his role as a behavioral health clinician and informatics director at Children's Hospital of Philadelphia.

Dr. Proctor's Journey [00:03:04]

Dr. Proctor shares his background, from his early interest in technology and psychology to his current role in informatics. He discusses his educational path and how he combined his passion for technology with his clinical work.

Informatics and Clinical Workflow [00:09:56]

Dr. Proctor explains the role of informatics in bridging the gap between clinical workflows and technology. He emphasizes the importance of understanding both the clinical and technical aspects to create effective solutions.

AI in Healthcare [00:11:23]

The discussion shifts to the implementation of AI in healthcare, specifically the use of large language models (LLMs) like ChatGPT in the EHR system. Dr. Proctor talks about the benefits and challenges of using AI to draft patient messages and reduce clinical burden.

Pilot and Implementation [00:13:30]

Dr. Proctor describes the pilot phase of implementing AI tools at CHOP and the transition to a full-scale rollout. He shares data on adoption rates and time savings, highlighting the differences in usage between physicians and nurses.

Prompt Engineering [00:14:48]

The conversation delves into prompt engineering, where Dr. Proctor explains how instructions are given to LLMs to guide their responses. He discusses the art of creating effective prompts and the challenges of adapting to changes in AI models.

Addressing Privacy Concerns [00:18:23]

Dr. Proctor addresses privacy concerns related to AI in healthcare, particularly in pediatric settings. He explains how CHOP ensures that sensitive information is protected when using AI tools.

Future of AI in Healthcare [00:27:30]

The hosts and Dr. Proctor speculate on the future of AI in healthcare. They discuss the potential for AI to handle more complex tasks, the integration of domain-specific models, and the importance of maintaining human oversight.

Nerd Alert Round [00:37:37]

The hosts engage Dr. Proctor in a rapid-fire round of questions, covering topics like his favorite efficiency hacks, home automation, and his interests outside of work.

Closing Remarks [00:49:51]

Dr. Proctor offers advice to early-career clinicians interested in informatics and technology. The hosts thank him for his insights and wrap up the episode.

In this episode of "NerdMDs Efficiency Unlocked," hosts Adam Carewe and Dale Gold interview Adam Cherrington, Vice President of Digital Health and Patient Voice at KLAS Research. They discuss the importance of patient engagement in healthcare, the role of technology in improving patient experiences, and the future of healthcare innovation.

Introduction [00:00:12]

The hosts introduce the episode and their guest, Adam Cherrington. They discuss the importance of improving patient experiences in healthcare and set the stage for the conversation.

Adam Cherrington's Journey [00:02:42]

Adam shares his journey into healthcare technology, his passion for patient engagement, and personal experiences that shaped his career. He highlights the importance of personalized care and accessibility in healthcare.

What is KLAS? [00:09:56]

Adam explains the mission and work of KLAS, an independent research and consulting firm dedicated to improving healthcare. He discusses the company's history, its role in providing transparency in healthcare, and its impact on the industry.

Digital Health and Patient Voice [00:14:01]

Adam delves into his specific role at KLAS, focusing on digital health and patient engagement. He discusses the challenges and opportunities in patient engagement, the importance of involving patients in their care, and the evolving landscape of healthcare technology.

Patient Voice Collaborative [00:22:52]

Adam introduces the Patient Voice Collaborative, a new initiative by KLAS to include patient perspectives in healthcare IT. He shares insights from early patient interviews and discusses the potential impact of this collaborative effort on healthcare organizations.

Future of Healthcare [00:35:35]

The hosts and Adam discuss the future of healthcare, including the importance of data interoperability, better patient-provider communication, and the need for consumer-friendly healthcare solutions. They share their hopes and dreams for a more integrated and efficient healthcare system.

Nerd Alert Round [00:40:05]

The episode concludes with a fun "Nerd Alert" round, where Adam Cherrington answers light-hearted questions about his personal preferences and interests, adding a touch of humor and personality to the conversation.

In this episode, Ruben Amarasingham discusses the latest advancements at Pieces, focusing on their innovative system to classify, detect, and prevent hallucinatory errors in clinical summarization. The conversation delves into the development and impact of their proprietary Safe Read platform, the challenges faced, the October publication they released to the public, and the future of AI in healthcare.

Section Summaries

Introduction [00:00:01]

Adam Carewe welcomes Ruben Amarasingham back to the podcast, highlighting the recurring guest status and setting the stage for a discussion on the latest developments at Pieces.

Overview of Pieces' Whitepaper [00:01:40]

Adam introduces the main topic of the episode: a whitepaper by Pieces titled "A system to classify, detect, and Prevent Hallucinatory Error in Clinical Summarization." He provides a high-level summary of the paper, emphasizing the importance of addressing hallucinations in AI-generated clinical summaries.

Reuben's Insights on the Whitepaper [00:04:20]

Reuben elaborates on the motivation behind the whitepaper, explaining the significance of creating accurate and reliable clinical summaries. He discusses the challenges of hallucinations in large language models and the framework Pieces developed to classify and mitigate these errors.

Development of the Classification Framework [00:07:36]

Ruben shares the history and development process of the classification framework, highlighting the importance of inter-rater reliability and the rigorous testing involved. He introduces the Safe Read platform, which helps measure and manage hallucinations in clinical summaries.

Safe Read Platform and Its Impact [00:11:23]

The conversation shifts to the Safe Read platform, an internal tool used by Pieces to detect and address hallucinations. Ruben explains how the platform works, its components, and the continuous improvement process through adversarial AI and knowledge graphs.

Challenges and Evolution of Safe Read [00:13:38]

Ruben discusses the challenges faced in developing and iterating the Safe Read platform. He highlights the importance of human intervention in improving the system and the role of adversarial AI in detecting hallucinations.

Impact of Safe Read on Clinical Summarization [00:18:46]

Ruben talks about the positive impact of the Safe Read platform on reducing hallucinations in clinical summaries. He explains how the system has evolved to address less severe hallucinations and improve the overall accuracy and reliability of summaries.

Industry Standards and Regulatory Challenges [00:22:19]

The discussion touches on the lack of industry benchmarks for AI systems in healthcare and the role of regulatory bodies. Ruben shares insights on Pieces' interactions with the Texas Attorney General and the importance of transparency and reporting in AI development.

Future Directions for Pieces [00:33:15]

Ruben outlines the future goals for Pieces, including expanding the Safe Read platform to other forms of AI documentation and patient interactions. He discusses the potential for collaboration with other AI companies and the broader impact on healthcare.

Closing Thoughts and Nerd Alert Questions [00:42:05]

Adam wraps up the conversation with some light-hearted "nerd alert" questions, asking Ruben about his favorite ways to relax, recent books he's read, and skills he wants to master. Ruben shares his love for reading, spending time with family, and his interest in art and AI-driven creative projects.