According to a 2019 study from the Journal of the American Medical Association, the U.S. healthcare system wastes an estimated $760 billion to $935 billion per year, which represents about 25% of total healthcare spending.
Medication-related problems carry a particularly heavy financial toll, with the total cost of non-optimized medications and the subsequent consequences estimated at $528 billion in 2016, or 16% of all U.S. healthcare expenditures.
Three of the largest areas of waste include failure of care delivery, failure of care coordination, and overtreatment or low-value care. Humans, though well-intentioned and capable, cannot keep up with the increasing administrative task complexity of a fractured healthcare delivery system.
To curb this waste, healthcare companies are turning to artificial intelligence technologies to streamline their processes while decreasing their spending. In the decades since its inception, artificial intelligence (AI), which includes all technologies that allow computers to simulate the human brain and complete human tasks, has generated interest in its possible applications in healthcare. Some of the applications of AI in healthcare include prevention, accurate diagnosis, treatment plans, and rehabilitation.
AI tools are a powerful way to scale medication management services and improve patient outcomes in a cost-efficient way. This article will review types of AI in healthcare and what impact AI can have specifically on medication optimization.
A rule-based expert system is a form of AI that uses facts and rules to solve a problem and deliver a solution. Rule-based AI systems are considered the simplest form of artificial intelligence. They mimic the skills of human experts using a set of IF-THEN rules. While this basic rule-based form of AI has many healthcare use cases, they have a limited scope because too many rules can make the system complicated and unreliable.
Most clinical decision support systems in healthcare use some form of rule-based expert systems. For example, if a patient is allergic to ibuprofen and has an order for aspirin, then the pharmacist verifying the order may get an alert suggesting that aspirin may be contraindicated.
Machine learning (ML) is a subset of AI that focuses on how computer systems learn from datasets and as a result, continuously improve their AI algorithms. The efficacy and accuracy of machine learning depend on the quality of the data the system learns from.
ML has been used to detect diseases early, particularly in cases where it’s difficult for a human clinician to do so, such as breast cancer or lung cancer. As machine learning algorithms analyze and learn from the data they ingest, they become more accurate over time.
One study looking at the impact of ML on a patient population of 43,000 individuals found that using ML to predict the risk of 30-day readmission could prevent 50% of readmissions and lower healthcare costs by $1 million.
Predictive analytics use patterns of information to predict future events, allowing early and targeted intervention and potential cost savings. In medication optimization, predictive analytics can forecast which medication-related problems patients are more likely to experience (e.g., medication non-adherence, drug interactions, or adverse effects) and facilitate pharmacist interventions that are specifically tailored to a patient’s risk profile. Healthcare organizations can use predictive analytics to identify patients who could benefit the most from clinical pharmacy interventions and assign the resources of medication therapy and disease management programs to these patients.
AI can also be used in healthcare for workflow and administration tasks. The use of AI in these applications could potentially save the healthcare industry $18 billion by decreasing the amount of non-patient care activities in clinician workflows and allowing organizations to optimize their resource allocations. These administrative applications can assist organizations in scheduling patients, prioritizing their team worklists, improving documentation, ensuring compliance, and detecting abuse and fraud.
Traditional medication management, because of its tedious and time-consuming processes, can only reach a restricted number of patients. Plus, pharmacists may have access to limited and often inaccurate patient data which makes it harder to manage patients’ medications effectively and comprehensively.
Implementing AI can transform medication optimization, allowing it to scale and improve cost savings by catching medication-related problems before they become costly, reducing mis-prescribing, and comprehensively improving adherence.
Predictive analytics allows health plans to predict the risk trajectory of their members and stratify them based on their risk for non-adherence or medication-related problems. With this ability to prioritize, health plans can assign their resources and focus their medication optimization programs on patients who need it the most.
Arine’s AI-powered risk stratification enabled one Arine client to significantly reduce hospitalizations, improve formulary adherence, and lower the cost of chronic disease management for its members by implementing data-driven interventions. Arine’s platform continuously measured the clinical and economic impact of clinical interventions. Then, using machine learning, Arine applied the insights through feedback loops and improved each intervention. As a result, the health plan achieved a 15% reduction in the total cost of care within 6 months and a 47% reduction in hospital readmissions among the members who received Arine’s interventions.
Leveraging an AI-based workflow engine in medication optimization processes streamlines team collaboration and prioritizes patients for intervention, assigning them to the appropriate member of the care team. Arine’s medication intelligence platform automatically assigns at-risk patients, creating tasks that guide care teams through the medication optimization process. This clinical workflow engine helps medical professionals stay hyper-focused on caring for patients who need them and who will see the most impact from medication optimization.
AI can also analyze a vast amount of healthcare data to identify prescribing patterns and outliers, triggering outreach to prescribers who need support and addressing medication-related problems at the prescriber level. Arine’s Prescriber Analytics solution leverages AI to intelligently identify and resolve prescribing aberrations, enabling health plans to send targeted education and recommendations to healthcare providers.
Using prescriber analytics, one Arine client reduced behavioral health polypharmacy by 45% to 55% for its members and achieved a 20% increase in adherence to behavioral health medications. The same client also lowered its members’ average daily morphine milligram equivalents (MME) by 20% and achieved savings in behavioral health-related spending of $1,500 to $4,300 per member per year.
Artificial intelligence can lead to more personalized care by integrating clinical, behavioral, and socio-economic data in patients’ care plans. Arine’s medication intelligence platform analyzes multi-source, multi-dimensional clinical data beyond just claims to drive both insights and action that enable healthcare professionals to build a therapeutic relationship, collect patient-generated medical data, and better assess the patient’s needs. The platform then uses all data to automatically create comprehensive customized care plans that are more likely to be adopted by members.
Using AI to scale services while personalizing care was crucial for one of Arine's clients in closing the racial gap that existed in their members’ medication adherence by 35% in 6 months.
In an ever-evolving and increasingly complex healthcare landscape, it’s essential for health plans to find new ways to improve their health outcomes and well-being and decrease their costs.
Applying artificial intelligence in healthcare can deliver outsized outcomes and enable health plans to scale their medication optimization programs in a cost-efficient way.