Nationwide Children’s Hospital has developed a machine-learning tool, the Deterioration Risk Index (DRI), that identifies children at risk of deterioration. The device, which was integrated into the electronic medical record, sounds an alarm and notifies the care team when a patient is at a high risk of deterioration. The DRI decreased degradation incidents by 77% one and a half years after its adoption. The DRI was successful because of collaboration and transparency. Other hospitals can retrain the algorithm on their data, improving care for children at their facility.
Artificial intelligence (AI) and machine learning (ML) have become ubiquitous terms in the world of technology. From chatbots to self-driving cars, AI and ML are being used to transform a variety of industries. Healthcare is no exception. There is an abundance of data that can be used to enhance patient care as a result of the growing digitization of patient data and the widespread use of electronic medical records (EMRs).
Machine learning is a type of AI that enables computers to learn from data without being explicitly programmed. Instead of following a set of rules, machine learning algorithms can detect patterns in data and use those patterns to make predictions or decisions. Many of the AI applications that we use daily, such as facial recognition, product recommendations, and spam filtering, are based on machine learning.
One of the most promising applications of AI and machine learning in healthcare is the ability to predict patient deterioration. When a patient’s condition deteriorates, their risk of morbidity or mortality increases. Early detection and intervention can be the difference between life and death. At Nationwide Children’s Hospital in Ohio, a team of experts in critical care, hospital medicine, data science, and informatics has developed a machine-learning tool that can identify children at risk for deterioration.
The Deterioration Risk Index (DRI) tool is trained on disease-specific groupings, including cancer, structural heart abnormalities, and general (neither cancer nor heart defect). The study team increased the tool’s accuracy by tuning the algorithm for each subpopulation. The likelihood that a patient will deteriorate is influenced by a variety of variables, including shifting lab values, medications, medical history, nursing observations, and more. As the DRI is integrated into the electronic medical record, the algorithm can use all the data and analyze it in real-time. It sounds an alarm if a patient becomes at high risk for deterioration, triggering the action and attention of the care team.
One of the key advantages of the DRI is that it is transparent. The team can show clinicians what data goes into the algorithm and how the algorithm evaluates it. The DRI team has also published the full algorithm in its report in the journal Pediatric Critical Care Medicine. Using this information, other hospitals can retrain the algorithm on their data to help improve care for children at their hospital.
The DRI has already had a significant impact on patient outcomes. A year and a half after the team implemented the tool, deterioration events were down 77% compared to expected rates. The tool was integrated into existing hospital emergency response workflows, and when an alert sounds, the care team responds with a patient assessment and huddles at the bedside to develop a risk mitigation and escalation plan for the identified patient.
While the DRI is a promising example of how AI and machine learning can be used in healthcare, it is not without its challenges. Many algorithms have been developed to predict risk and improve clinical outcomes, but the majority don’t make it from the computer to the clinic. According to the DRI team, collaboration and transparency were key to making the DRI work in the real world. The tool was in development for more than five years, during which time the team met with clinical units and demonstrated the tool in its various stages of development. In those meetings, the care teams asked questions and provided feedback.
AI and machine learning is being used in a variety of other applications in healthcare as well. For example, AI-powered chatbots can be used to triage patients and provide guidance on when to seek medical attention. AI is also useful for improving diagnostic accuracy by analyzing medical images, such as X-rays, CT scans, and MRIs. In one study, a deep learning algorithm was able to identify breast cancer with greater.
