The Healthcare Information and Management Systems Society (HIMSS) conference is one of the most anticipated events in the healthcare industry. HIMSS 2023 was held in Chicago from April 17-21 and brought together healthcare leaders, vendors, and professionals to discuss the latest trends and innovations in HealthTech. Since this newsletter is about Health Data Science, here are three key takeaways from the HIMSS conference on Artificial Intelligence and Machine Learning.
Using Artificial Intelligence To Find People With Rare Diseases.
Artificial Intelligence (AI) was a hot topic at HIMSS 2023. One of the presentations discussed how AI can be used for detecting rare diseases. Danita Kiser, Ph.D., used an example involving Sickle Cell Disease and Beta-thalassemia. The presenter suggested using deep learning models to help predict rare diseases based on the patient’s medical history or inpatient event data. The example she used involved using healthcare claims data dating back to 1993; she had data for 101 million people. The presenter concluded her presentation by suggesting that applying deep learning to rare disease detection would enable earlier treatment of rare diseases due to earlier detection.
Improving Nurse Staffing Schedules Using Machine Learning
Efficient nurse staffing is crucial for providing high-quality patient care and achieving positive health outcomes. However, creating and maintaining optimal nurse staffing schedules can be challenging and time-consuming. This year at HIMSS, there was a presentation about this topic from the Indiana University Health System. The data they used for the models was historical hospital data, patient order data, patient census data, acuity levels, ICD10 codes, and day-of-the-week data. In addition, they incorporated external data like weather data and historical flu/pandemic data. After setting up their end-to-end FHIR pipeline in GCP, a dashboard was created in Looker. Then, eight machine learning models were developed to predict demand, and they implemented a shadow dynamic scheduler to account for variations in the existing Kronos static scheduling method. Next, the dashboard was updated with predictions from the ML model and retrained as needed. Eventually, they were able to do implement the models across the organization.
Machine Learning Used to Reduce Medication Errors
Medication errors can have severe consequences for patients and lead to adverse drug reactions, hospitalization, and even death. This year at HIMSS, there was a presentation on how machine learning can reduce medication errors. First, the presentation reviewed the current state of medication errors and the process for building machine-learning models to detect discharge medication errors. For the modeling process, they have a ground truth dataset. Then, they trained the machine learning models and developed an error prediction pipeline. Last, they deployed the results to a dashboard and integrated that into real-time decision support. The dashboard they shared during the presentation included a visualization of total errors by race, hospital department/service errors, and a moving average chart of the medication errors.
Bonus
Some sessions were on prioritizing maternal health to advance interoperability and reduce mortality. Great sessions. I recommend following some of the speakers, Susan Kressly, MD, Hannah Galvin, MD, Nakyda Dean, MD, Kate Drone, Asha Emmanuelle, RN, and Evelyn Gallego, MBA/MPH.
In conclusion, HIMSS 2023 highlighted the growing role of data science and machine learning in improving healthcare. Healthcare organizations are embracing data science to improve patient outcomes, increase efficiency, and reduce costs. As the healthcare industry continues to evolve, it is evident that data science and machine learning will play a critical role in shaping the future of healthcare.