Sam Lapiejko ’26 stands in the front of a Unistructure classroom chatting with peers about his poster on automated rotator cuff tear classification using 3D convolutional neural networks. Walking individuals through the academic paper he chose for his literature review, Lapiejko shares that the study’s authors used a computer program to look at 3D images of shoulders to locate tears and show how bad they were. Using the paper as a guide, Lapiejko then used comparable techniques on similar healthcare images to see if artificial intelligence tools could accurately detect tears.
Lapiejko is one of 14 students in Biological and Biomedical Sciences’ Professor Brian Blais Ph.D.’s “AI in Healthcare” course. A new offering within the School of Health and Behavioral Sciences, this course focuses on applied techniques in AI. Throughout the semester, students gain skills in implementing methods of AI in healthcare as well as an understanding of how these methods work.
“AI is becoming more prevalent and changing every day — you’ve got to get ahead of the curve,” says Lapiejko, a Finance major whose interest in the groundbreaking technology drew him to Blais’s course.
An evolving landscape
Beginning with the basics, Blais starts the semester by sharing papers on current healthcare advancements that use AI. He then walks undergrads through the content to ensure they understand the literature and to help them build their vocabulary, which is broken into medical vocabulary, AI vocabulary, and mathematical vocabulary. Together, they define the process that researchers are using, consider why the studies are using certain data methods, and discuss potential ethical and privacy issues related to healthcare.
“Healthcare and AI is something of general interest, and these topics impact you personally no matter who you are,” says Blais, adding that his students come from a variety of disciplines.
RELATED ARTICLE: Stuck in an exercise rut? AI can help, says Bryant expert
He notes that understanding how AI works within healthcare takes away some of the mystery associated with it.
“There is a direct algorithm — the AI is not pulling things out of thin air,” Blais says. “The more you think of it as magic, the less connected you will be to the technology.”
Poster demonstrations throughout the semester help students apply what they’ve learned by explaining the research and processes to others in lay terms. By the end of the semester, students are able to approach new methods of AI with confidence as they understand AI’s strengths and weaknesses.
Healthcare of the future
To wrap up the semester, students hold another demo — this time focusing on large language models, which included several types of applications, such as predicting patient outcomes using electronic medical records, clinical trial matching, and the use of Chat-like applications in research and clinical work.
“All of this is becoming more common, with essentially doctor ‘plus’ AI being better than either doctor alone or AI alone,” Blais says.
He adds that this portion of the course offered curveballs for students as they encountered challenges with the availability of data.
“Privacy issues are forefront here, and it was hard to find good data sets to learn from. This brought us into another part of the field which is the use of synthetic data – essentially data that is made in the form of real data but doesn’t include actual patient data,” Blais says, emphasizing that this was a productive area of investigation since it enabled more efficient research and teaching of the subject.
Andrew Cumming ’25, reflecting on his classroom experience, notes that Blais’s course helped him understand new ways to harness the power of AI. From his image recognition presentation on using AI to locate areas that are prone to lung cancer to his large language model presentation on predicting lung cancer patient prognosis with LLMs, the Accounting major developed a new understanding of the technology’s current, and future, impact.
“Some cancers are harder to detect than others and improving models like this can help everyone in the long run,” says Cummings.
