AI For Clinical Decision Support

PCCI’s clinical decision support AI/ML predictive models are designed to empower healthcare teams by providing augmented intelligence and insights to generate early warnings for clinical deterioration, enhance their point-of-care decision-making, and enable them to intervene earlier and with more precision.

PCCI’s Clinical Decision Support AI/ML Models

AI In Action: Parkland Trauma Index of Mortality (PTIM) for Real-Time Trauma Mortality Prediction

The Parkland Trauma Index of Mortality (PTIM) is a machine learning algorithm and the only known model that uses electronic medical record data to predict—every hour—48-hour mortality during the first 72 hours of hospitalization, thus evolving with the patient’s physiologic response to trauma. Over a one-year period, PTIM has correctly identified 89% of the high-risk trauma patients and 92% of the low-risk trauma patients.

This model was developed because provider teams needed a dynamic tool to estimate patient mortality risk for prioritization of interventions as the teams must make many consequential decisions about trauma care in rapid succession; i.e., when to stabilize versus intervene, how to sequence interventions, etc. In many cases, these decisions can feel like equal parts instinct, art, and science, leading to delays and disagreements among care teams.

PTIM has the potential to make an enormous impact on trauma centers to enhance patient care, especially in the first 72 hours, when many of the critical-condition patients (from motor vehicle accidents or other traumatic events) arrive at hospitals with no identification and little to no information about their past medical histories. Prior industry models provided only a static, one-time score at the time of admission.

“Over a one-year period, PTIM has correctly identified 89% of the high-risk trauma patients and 92% of the low-risk trauma patients.“

PCCI’s AI/ML Model to Predict HIV Infection Risk

PCCI is also collaborating with DCHHS and Parkland in a Dallas County public health initiative targeted to adults who are at risk for exposure to HIV through an HIV-STI pre-exposure prophylaxis (PrEP) Initiative.

Dallas County ranks 2nd highest in HIV, 6th in Syphilis, 21st in Gonorrhea, and 26th in Chlamydia infection rates compared to the other 254 Texas counties. PCCI built a successful HIV ML prediction model using a Light Gradient Boosting Machine (LGBM) algorithm. LGBM is an ensemble of decision trees trained sequentially one after the other, improving from the errors of the predecessor to result in a strong boosting classifier.

Overall, this model uses 26 input variables to predict the individuals at increased likelihood of acquiring HIV and who are candidates for HIV PrEP, with the overall goal to both increase awareness and decrease rates of Sexually Transmitted Infections (STIs) (including HIV), to promote public health in Dallas County’s most affected ZIP Codes. This program and its results were published in the peer-reviewed publication, The Journal of Acquired Immune Deficiency Syndrome (JAIDS), co-authored by Arun Nethi, Data & Applied Scientist, and Albert Karam, MS, MBA, VP, Data Strategy Analytics.