Researchers at the University of Illinois Chicago have developed an elective course that can quickly transform fourth-year medical students from functional AI novices to budding AI experts.

Trying the four-week curriculum on two separate cohorts, the team found participants boosted average AI knowledge scores from 61% pre-course to 97% post.

Describing the project in the July edition of the Journal of the American Medical Informatics Association, corresponding author Jacob Krive, PhD, and colleagues suggest four competencies that medical-school faculty could instill in their students with similar success:

Competency 1. Evaluator. Key skill: Ability to discern when a technology is appropriate for a given clinical context and what inputs are required for meaningful results. Learning objectives:

• Explain the foundational concepts of medical analytics AI and explainable AI.
• Describe a clinical scenario in which an AI solution can augment and improve clinical processes.

Competency 2. Critical AI systems appraiser. Key skill: Ability to explain how various forms of artificial intelligence are applied today in healthcare to augment and improve health outcomes. Learning objectives:

• Build and appraise use cases of applied AI in value-based care being used by health systems today.

Competency 3. Interpreter. Key skill: Ability to assess AI inputs and outputs with a reasonable degree of accuracy, including knowing potential sources of error, bias or clinical inappropriateness. Learning objectives:

• Interpret common AI terms and components (algorithms, machine learning vs. deep learning, etc.)
• Define and describe linkages between evidence-based medicine, real-world evidence, medication safety, predictive analytics, mobile computing, artificial intelligence.

Competency 4. Communicator. Key skill: Ability to convey results and underlying process in a way that patients as well as other health professionals can understand. Learning objectives:

• Discuss leading practices in data integration needed to deploy analytic solutions in healthcare.
• Identify the drivers, decision factors and collaboration required across clinical, operational and technology teams to realize return on investment in AI.

Krive and co-authors note that, post-course, participating students demonstrated a good grasp of ways their fast-gained understanding might apply during upcoming residencies and, by extension, over the course of their respective careers in medicine.

The authors’ concluding comment:

“Education regarding the incorporation of AI technologies in routine clinical care, if well done, will give the gift of time to patients and physicians. This time will enable physicians to foster meaningful and therapeutic relationships while utilizing AI technologies to ensure the provision of optimal care and completion of more mundane tasks.”

JAMIA has posted the paper in full for free.

Buzzworthy developments of the past few days.

• AI had a hand in almost 4,000 job losses in May. So reports the outplacement firm Challenger, Gray & Christmas. Healthcare wasn’t hit especially hard. The tech, retail, automotive and financial sectors took bigger slams from AI and all other causes. Still, this latest report marks the first time Challenger has included AI in its employment survey, suggesting things not just to come but already here—for pretty much every industry. Report here, brief analysis by CBS News’s MoneyWatch here.
   
• Boards of directors at organizations developing or using AI need to ‘step up’ to head off confirmation bias. So believes Bill Schmarzo, data science instructor, influencer and wearer of many hats under the Big Data tent. Making the case at Data Science Central, Schmarzo writes that such data deficiency “is of significant concern in fields such as employment, education, housing, healthcare, finance and criminal justice, where the consequences of incorrect predictions can be destructive.” Read the post.
   
• GE HealthCare has won FDA clearance for its Sonic DL system. The letters stand for Deep Learning, and the company says the system dramatically speeds MRI image acquisition: It can capture high-quality cardiac images in a single heartbeat. Announcement.
   
• AI has eyes for blockchain. And the admiration seems mutual. As AI assistants are increasingly dispatched to help people navigate blockchain environments, blockchain becomes more and more accessible to non-experts. Cryptocurrency aficionado John Kiguru of Crypto News Flash breaks it down.
   
• Aesop Technology (Chicago) is working with AstraZeneca Taiwan to use AI for aiding clinicians treating cancer patients. Called Medigator, the software works by heading off unwanted disruptions that would have set back administration of immunotherapy drugs. Announcement.
   
• Researchers are using AI to improve existing techniques for fighting superbugs. The work is ongoing at the University of Michigan. The team says its model is simply the best at predicting interactions between proteins and nanoparticles, a key tactic in the war against antibiotic-resistant infections. Details.
   
• The European Union is pressing social media companies to clearly label content created with a hand from generative AI. The stepped-up campaign reflects the EU’s determination to fight fake news and disinformation, not least hogwash emanating from Russia as part of its war strategy against Ukraine. (Not that all sides don’t do it.) Coverage current in The Guardian, NBC News, Associated Press, Reuters, ABC News …
   
• Patients undergoing painful medical procedures may soon choose to let virtual reality lull them into a state of deep relaxation. A study supporting the hypnotic technique’s merit is running in a peer-reviewed journal and summarized in Health Imaging.