HMN 2025: How AI can forecast COVID-19 dangers and therapy for hospitalized sufferers

Seasonal influenza, respiratory syncytial virus (RSV), and COVID-19 are actively circulating all through the United States. These respiratory sicknesses are contributing to widespread well being considerations, with circumstances being reported in numerous areas nationwide.

Using synthetic intelligence and machine studying, researchers from Florida Atlantic University’s Christine E. Lynn College of Nursing and College of Engineering and Computer Science, in collaboration with Memorial Healthcare System, are pushing the boundaries in well being care to foster innovation, improve decision-making, and finally enhance well being outcomes for people and populations.

To predict the severity of COVID-19 and finest therapeutic interventions through the pandemic, researchers established an AI-driven resolution help system by figuring out essential options influencing the severity of illness outcomes in sufferers hospitalized with COVID-19 in a South Florida hospital.

Specifically, the research targeted on predicting the necessity for intensive care unit (ICU) admission with or with out mechanical air flow and intermediate care unit (IMCU) admission. The purpose was to leverage these options to allow sooner and extra correct forecasting of therapy plans, probably stopping essential situations from worsening.

For the research, researchers analyzed digital well being report (eHR) information from 5,371 sufferers admitted to a South Florida hospital with COVID-19 between March 2020 and January 2021. They educated three Random Forest fashions to foretell mechanical air flow, ICU, and IMCU admission utilizing 24 variables, together with sociodemographics, comorbidities, and drugs. The evaluation centered on information collected on the time of hospital admission.

Results of the research, published within the journal Diagnostics, in early fall 2024, present that the fashions for ICU with mechanical air flow, ICU, and IMCU admission recognized the next components overlapping as an important predictors among the many three outcomes: age, race, intercourse, physique mass index (BMI), diarrhea, diabetes, hypertension, early phases of kidney illness, and pneumonia.

Researchers additionally discovered that people 65 and older (“older adults”), males, present people who smoke, and BMI categorised as “obese” and “overweight” had been at better danger of severity of sickness. The research additionally explored the severity of the illness below the co-occurrence of danger components.

“This is among the only a few research that explored such interactions amongst danger components utilizing machine studying interpretability approaches. For instance, pneumonia mixed with diabetes elevated mechanical air flow danger, whereas diarrhea interacted strongly with diabetes for ICU admissions,” stated Debarshi Datta, Ph.D., senior writer and an assistant professor in FAU’s Christine E. Lynn College of Nursing.

“IMCU severity was linked to combos like diarrhea with pneumonia and hypertension in older adults. Additionally, drugs equivalent to angiotensin II receptor blockers and ACE inhibitors appeared to decrease illness severity, aligning with prior analysis on their protecting results.”

The high options recognized by the fashions’ interpretability had been from the “sociodemographic traits,” “pre-hospital comorbidities,” and “drugs” classes. However, “pre-hospital comorbidities” performed an important function in several essential situations. In addition to particular person function significance, the function interactions additionally present essential info for predicting the more than likely final result of sufferers’ situations when pressing therapy plans are wanted through the surge of sufferers through the pandemic.

Compared to earlier research, this novel method stands out through the use of readily accessible eHR information and mixing machine studying interpretability strategies with conventional statistical strategies. This technique enabled a deeper understanding of options like age, intercourse, BMI, and comorbidities equivalent to diabetes and hypertension throughout totally different severity ranges.

“While biomarkers have been utilized in different research, their restricted scientific accessibility makes our findings extra sensible for real-world well being care functions,” stated David Newman, Ph.D., co-author, professor, and statistician, FAU Christine E. Lynn College of Nursing.

“By figuring out essential components and interactions influencing COVID-19 outcomes, our research offers actionable insights for bettering affected person care and supporting well being care methods throughout high-demand situations.”

Importantly, the appliance of AI/machine studying in well being care extends past the COVID-19 illness, holding promise for bettering prognosis, therapy choice, illness surveillance, and affected person outcomes throughout numerous medical specialties and well being care settings. This information empowers public well being authorities to proactively plan and implement focused interventions, mitigating the impression of illness outbreaks and optimizing well being care supply.

“Developing an AI-driven resolution help system to foretell essential scientific occasions in COVID-19 in-patients not solely meets the pressing calls for of a pandemic but in addition breaks new floor in AI and machine studying in well being care,” stated Datta.

“By using superior applied sciences and algorithms, equivalent to machine studying, researchers and clinicians can harness the facility of data-driven insights to revolutionize affected person care.”

Study co-authors are Subhosit Ray, Ph.D., a postdoctoral fellow; Laurie Martinez, Ph.D., an assistant professor; Safiya George Dalmida, Ph.D., former dean; all with FAU’s Christine E. Lynn College of Nursing; Javad Hashemi, Ph.D., inaugural chair and professor of the Department of Biomedical Engineering and affiliate dean for analysis, FAU College of Engineering and Computer Science; Candice Sareli, M.D., vice chairman and chief medical analysis officer, Memorial Healthcare System; and Paul Eckardt, M.D., chief, Memorial Division of Infectious Disease, Memorial Healthcare System.