AI and Robotics at TRU: Institutional Augmentation – Part 4

“Prior to going to clinical, [students are] that much more prepared for working with a real patient,” TRU assistant teaching professor Pinder Nagra said. “If they respond to something incorrectly here, we believe that’s a great learning opportunity and they can use that to prevent that from happening in real-world practise.”

In Episode 4 of AI and Robotics at TRU: Institutional Augmentation, Nagra and Yasin Mamatjan, chair of the TRU engineering department, spoke to CFJC Today about some of the technology that his helping their research and teaching.

Nagra sat next to a computer for a show-and-tell session on Simulated Client Records, a system that won an Open Education Award for Excellence and was developed by Nagra, Devon Graham and TRU Open Press.

The free online platform for educators and students is designed to encourage familiarity with electronic health records and medical charts, the ones used when split-second decisions are of great consequence.

“To decrease that anxiety and also help them navigate what that chart would look like in real life, so they’re developing their clinical reasoning by doing so,” Nagra said.

Clinical reasoning is also developing through another TRU innovation, AI SimSpeak, an open access website that features pre-programmed chatbots at varying stages of life, each with medical issues to explore.

“We’ve even programmed them to be very specific, so she might use slang that a typical 15-year-old would use,” said Nagra, who worked with Anila Virani and Brad Forsyth on the project, with help from a CAN-Sim Grant. “Perhaps [they’re having] a mental-health crisis. Their answers would be similar to that of someone experiencing that.”

The virtual clients are a source for simulated interaction.

“[Students are] developing their communication skills and in doing so, they’re also able to navigate when there is an issue, how they would respond to it and then create healthcare plans, health promotion plans for that individual patient,” Nagra said. “We’ve also tried to create patients who are diverse to exemplify our community within Kamloops.”

Nagra, with the help of a simulation tech, slid on a pair of virtual-reality goggles that offer a window into the world of patients who have dementia.

Users can navigate through an apartment while struggling with colour perception and experiencing tinnitus and photo sensitivity.

“Confusion, anger, their inability to navigate their surroundings, which causes frustration – students were all put in that position and quickly realized how difficult that would be,” Nagra said.

Mamatjan, who runs the AI for Digital Medicine Lab, is developing potentially life-saving biomedical devices that measure vital signs such as blood pressure and heartbeat to assess risk factors that can lead to strokes.

“We build the hardware tools that continuously monitor some high-risk people,” Mamatjan said. “If risk factors increase, then it automatically gives warnings. At TRU, my group has developed systems for stroke risk stratification and multimodal monitoring, integrating signals such as physiological data, patient movement, image and vital signs.”

He said the tools include real-time prediction models and intelligent monitoring designed for earlier detection and better decision support. They come equipped with a chatbot.

“It can send a message and also even is able to send messages to their relatives, as well,” Mamatjan said.

The AIdMed laboratory integrates artificial intelligence, software engineering and biomedical research to tackle critical challenges in healthcare, Mamatjan said, noting the lab’s focus is to create AI-powered tools for patient monitoring and health coaching.

Another significant area of Mamatjan’s work is improving cancer diagnosis through molecular AI models.

“Tumour classification in the clinic is primarily based on histo-morphologic analysis, but tumor types may have indistinguishable features under the microscope and there are problems (with inter-observer variability) in histology-based classification,” he said. “However, molecular profiling is performed for cancer genomics analysis such as mRNA and DNA methylation. AI models then integrate pathology and molecular data to classify the tumor types and subtypes more accurately and support treatment decisions. We wanted to show that genomics can be helpful in characterizing these tumours.”

by Marty Hastings

AI and Robotics at TRU: Institutional Augmentation – Part 4

Our Community