Generative AI is the star of HIMSS 2024 in Orlando and GE HealthCare added to the hype this week, unveiling several initiatives and partnerships that leverage the innovative technology.
The healthcare industry is still at the beginning stages of utilizing generative AI to the fullest extent but progress has been made. So far, it’s been used to streamline efficiency across different areas of the healthcare sector, including administrative tasks in the clinical space.
However, in the words of Parminder Bhatia, the chief AI officer at GE HealthCare, the technology has had more of an impact in the last 18 months than it has in the last 10 years.
“Initial use cases with GenAI will focus on operational efficiency, and how to reduce the time on mundane tasks,” Bhatia told MM+M. “A lot of focus [at HIMSS] will be around doing that responsibly.”
As large language models have expanded over the past couple of years, Bhatia envisions 2024 being the moment for large multimodal models to take the next step.
“Healthcare is truly multimodal and that’s why some of these technologies will work better integrated now than in the past,” Bhatia said. “It’s tailor made for healthcare. How can data from radiology, pathology and clinical notes get combined? There are a lot of exciting discussions [at HIMSS] and I’m looking forward to having some synergies with our partners and customers around that as well.”
GE’s GenAI offerings
At HIMSS, GE HealthCare is highlighting several of its new AI-enabled medical devices and programs — many of which aim to alleviate clinical workloads and make tasks more efficient.
One of its latest tools, CardioVisio, is a digital mechanism that helps clinicians analyze longitudinal data to monitor disease progression in atrial fibrillation (Afib), a form of irregular heartbeat. CardioViso can provide clinical decision support based on up-to-date Afib guidelines, Bhatia said.
GE Healthcare also touted its Command Center Software Platform, which integrates streaming data from electronic medical records and provides insights to caregivers through “Tiles,” or decision support applications.
Then there’s Portrait Mobile Wireless and Wearable Monitoring Solution, which allows real-time monitoring of a patient’s vitals to help track critical recovery, such as after surgery or following discharg from the ICU.
GE HealthCare also promoted its Vscan Air and Viscan Digital Tools, which are handheld ultrasound tools that can provide quick cardiac assessments and more.
“HIMSSis a good forum as the year starts to hear how the industry is evolving, and how different partners and ecosystems are looking to think about AI from various perspectives,” Bhatia said. “As some of these technologies get built out, we envision a lot of three- to four-way partnerships where med tech companies like us [team up with] hyperscalers, clinical partners and research institutes to make these AI capabilities primetime ready.”
Provider partnership launched
GE HealthCare simultaneously announced a new AI partnership with Mass General Brigham that will boost the integration of medical imaging foundation models into AI research work.
The focus of the collaboration is to develop foundation models, guided by “responsible AI,” Bhatia said.
“To bring some of these technologies into the clinical care area, they will have to have the competence of responsible AI,” Bhatia said. “That means minimizing bias, making sure the models are robust and ensuring they have all the important guardrails to make them primetime ready.”
One of the challenges Bhatia hopes to see addressed by genAI is how to change the current clinical workflow in the smoothest way possible.
That requires training healthcare professionals to integrate AI “seamlessly” into electronic health records and ensuring AI complements human decision-making, he explained.
“The other aspect is that AI is not just for the output or the prediction — it can be used to input or combine the data that is coming from different sources,” he added. “There needs to be that level of understanding, and making sure there is transparency and guardrails. As long as you adhere to those, it becomes an easier proposition to bring some of these capabilities into the workflow.”