🚨 Pulse of Innovation 🚨

Breaking news in the healthcare AI

We’ll be honest: there was some exciting news last week, but nothing that was considered breaking news!

• Google partners with Bayer to offer radiologists AI tools. The platform can flag AI problems within images for radiologists to look at, emphasizing that the radiologist will have “sole control.”

• DrugGPT is a tool developed to help physicians prescribe medications.

• Kaiser Permanente to use AI to help redirect simple messages from physician” inbox”

🧑🏼‍🔬 Bench to Bedside👨🏽‍🔬

Developments in healthcare AI research and innovations

Health Care Cost Reductions with Machine Learning–Directed Evaluations during Radiation Therapy

• UCSF clinicians and scientists investigate the impact of machine learning (ML) on managing high-risk patients undergoing radiation therapy and the potential to enhance patient monitoring while reducing cost. Patients identified by ML as high-risk and receiving twice-weekly evaluations experienced fewer complications requiring acute care. This proactive approach allowed clinicians to address issues promptly, enhancing patient outcomes. Integrating ML into clinical evaluations could help healthcare providers prioritize resources and interventions more effectively.

• The economic analysis revealed that ML-guided protocols reduced the average medical costs by approximately $1616 per patient treatment course. These cost savings primarily resulted from decreased emergency visits and hospitalizations, showcasing the potential of ML to make healthcare delivery more cost-effective.

• While the study highlights significant benefits, the adaptability and generalizability of economic analysis is unclear. Additionally, the initial costs and infrastructure required to implement such ML systems remain to be seen.

👩‍⚕️AI in the Clinic 🏥

Real-world and practical use of healthcare AI in clinic

FDA authorizes Prenosis’ AI-based tool for sepsis diagnosis (read before you act)

• The Sepsis ImmunoScore is AI/Machine Learning (AI/ML) Software that identifies patients at risk for having or developing sepsis. It uses up to 22 predetermined inputs from the patient’s electronic health record to generate a risk score and assign the patient to one of four discrete risk stratification categories based on the increasing risk of sepsis.

• The Sepsis ImmunoScore is intended to be used in conjunction with other laboratory findings and clinical assessments to aid in the risk assessment for the presence of or progression to sepsis within 24 hours of patient assessment. It is intended to be used for patients admitted to the Emergency Department or hospital for whom sepsis is suspected, and a blood culture was ordered as part of the evaluation for sepsis. It should not be used as the sole basis to determine the presence of sepsis or the risk of developing sepsis within 24 hours.

• The challenges of such technology include the sensitivity/specificity of detecting sepsis. The pivotal studies were done at two centers with 1400 patients in the emergency department, and the receiver operating curve for diagnosing sepsis was 0.83. A score of 0.83 is good—closer, e.g., 0.5 means it's not much better than flipping a coin, and 1.0 means it's always correct.

🩺 Start-Up Stethoscope 💵 

Trending startups and technologies impacting clinical practice

ThinkSono: AI software to help augment Ultrasound use in clinical practice.

• ThinkSono was founded in London in 2016 by Fouad Al Noor and Sven Mischkewitz. Their AI-guided compression sonography for DVT diagnosis has demonstrated high rates of diagnostic quality scans and allows for 100% sensitivity for DVT triage. The study published in Annals of Vascular Surgery shows that 53.42% of formal ultrasound examinations could potentially be avoided with this protocol.

• ThinkSono Guidance, an AI App that pairs with Clarius ultrasound scanners, recently attained Class llb CE mark regulatory approval. ThinkSono Guidance enables non-ultrasound-trained healthcare professionals to use the Clarius handheld ultrasound scanner to collect DVT exam data for a qualified clinician to review. This can speed up the diagnosis from several hours to as little as 15 minutes.

🤖Patients First, AI Second🤖

Ethical and regulatory landscape of healthcare AI

The algorithm journey map: a tangible approach to implementing AI solutions in healthcare

• How do you implement an AI solution safely and ethically in your healthcare system? This is a growing need, and researchers at Duke University explain how they incorporated the SepsisWatch tool. The "Algorithm Journey Map" is a crucial tool that maps the roles and stages involved in the AI solution's lifecycle. If you are a healthcare system, absolutely try to see how this example can be a framework for integrating AI tools in the clinic.

• Clinicians and healthcare administrators should focus on continuous training and engagement with AI tools, ensuring that the healthcare staff are well-prepared and the AI systems are appropriately integrated within existing clinical workflows. Regular feedback loops and updates are essential to adapt to clinical needs and technological advancements.