While many healthcare organizations have implemented Artificial Intelligence (AI) and Machine Learning (ML) tools at the point of care, few have successfully applied them to high-level decision making. A new frontier is expanding AI from artificial intelligence to augmented intelligence; traditional AI focuses on improving analytics efficiency while augmented intelligence is about improving the decision-making ability of healthcare leaders. This article addresses the capabilities health systems should embrace and provides two examples of how AI can assist with leaders with their most important decisions. Healthcare leaders’ biggest needs of from AI are the ability to separate signal from noise and make decisions that impact the future.
Choose by Topic
- Latest Insights
- Clinician Engagement
- Data: Quality, Management, Governance
- Decision Support
- Enterprise Data Warehouse / Data Operating system
- Financial Alignment and ROI
- Health Catalyst Products and Services
- Healthcare Analytics Summit
- Leadership, Culture, Governance, Diversity and Inclusion
- Life Sciences
- Outcomes Improvement
- Patient Experience, Engagement, Satisfaction
- Patient Safety
- Payers and Providers
- Population Health and Care Management
- Regulatory Measures
- Value-Based Care
- Weekly News Roundup
AI - Additional Content
Spend time reading content for you
Machine Learning in Healthcare: What C-Suite Executives Must Know to Use it Effectively in Their Organizations
Machine learning (ML) is gaining in popularity throughout healthcare. ML’s far-reaching benefits, from automating routine clinical tasks to providing visibility into which appointments are likely to no-show, make it a must-have in an industry that’s hyper focused on improving patient and operational outcomes. This executive report—co-written by Microsoft Worldwide Health and Health Catalyst—is a basic guide to training machine learning algorithms and applying machine learning models to clinical and operational use case. This report shares practical, proven techniques healthcare organizations can use to improve their performance on a range of issues.
Health Catalyst President of Technology, Dale Sanders, gives straightforward answers to tough questions about the future of AI in healthcare. He starts by debunking a common belief: We are awash in valuable data in healthcare as a consequence of EHR adoption. The truth involves a need for deeper data about a patient.
Healthcare organizations that leverage both text analytics and machine learning are better positioned to improve patient outcomes. Used in tandem, text analytics and machine learning can significantly improve the accuracy of risk scores, used widely in healthcare to help clinicians identify patients at high risk for certain conditions and, therefore, intervene. Health systems can run machine learning models with input from text analytics to provide tailored risk predictions on both unstructured and structured data. The result? More accurate risk scores and the ability to identify every patient’s level of risk in time to inform decisions about their care.
In today’s improvement-driven healthcare environment, organizations must ensure that improvement measures help them reach desired outcomes and focus on the opportunities with optimal ROI. With data science-based analysis, health systems leverage machine learning to determine if improvement measures align with specific outcomes and avoid the risk and cost of carrying out interventions that are unlikely to support their goals. There are four essential reasons that insights from data science help health systems implement and sustain improvement:
- Measures aligned with desired outcomes drive improvement.
- Improvement teams focus on processes they can impact.
- Outcome-specific interventions might impact other outcomes.
- Identifies opportunities with optimal ROI.
Precision medicine, defined as a new model of patient-powered research that will give clinicians the ability to select the best treatment for an individual patient, holds the key that will allow health IT to merge advances in genomics research with new methods for managing and analyzing large data sets. This will accelerate research and biomedical discoveries. However, clinical improvements are often designed to reduce variation. So, how do systems balance tailoring medicine to each patient with standardizing care? The answer is precise registries. For example, using registries that can account for the most accurate, specific patients and disease, clinicians can use gene variant knowledge bases to provide personalized care.
The Dangers of Commoditized Machine Learning in Healthcare: 5 Key Differentiators that Lead to Success
Many vendors deliver machine learning models with different applications in healthcare. But they don’t all deliver accurate models that are easy to implement, targeted to a specific use case, connected to actionable interventions, and surrounded by a machine learning community and support team with extensive, exclusive healthcare experience. These machine learning qualities are possible only through a machine learning model delivered by a vendor with a unique set of capabilities. There are five differentiators behind effective machine learning models and vendors:
- Vendor’s expertise and exclusive focus on healthcare.
- Machine learning model’s access to extensive data sources.
- Machine learning model’s ease of implementation.
- Machine learning model’s interpretability and buy-in.
- Machine learning model’s conformance with privacy standards.
Introducing Touchstone: The Next-Generation Healthcare Benchmarking and Opportunity Prioritization Tool
To do healthcare benchmarking effectively and efficiently, healthcare organizations need to know where they’re underperforming, where they’re performing well, and how to focus and prioritize their improvement efforts. They also need a new approach to benchmarking that isn’t limited to the inpatient setting. The Health Catalyst® Touchstone™ product is the next-generation healthcare benchmarking and prioritization tool that delivers what antiquated benchmarking technologies cannot:
- Risk-adjusted benchmarking across the full continuum of care.
- Artificial intelligence-powered recommendations.
- Ranked lists of improvement opportunities.
- Detailed analytics and an intuitive user interface that enable the easy exploration of factors driving performance issues.
- Democratized benchmarking that’s available to as many people as the organization wants.
Healthcare leaders who understand data science can embrace the significant improvement potential of the industry’s vast data stores, including an estimated $300 billion in annual costs savings. Executives must know the value of data science to understand the urgency in investing and supporting the technology and data scientists to fully leverage data’s capabilities. Today’s data science-savvy executives will lead the healthcare transformation by enabling faster, more accurate diagnoses and more effective, lower-risk treatments.
The FDA recently released guidance documents on the use of clinical decision support (CDS) and medical software that may be of concern to forward-thinking healthcare innovators who rely on these technologies to deliver exceptional care and improve outcomes. What will be the impact of this guidance on machine learning and predictive analytics efforts? How will the guidance affect timelines, costs, and effectiveness of ongoing machine learning implementation? As healthcare delivery increasingly relies on digital innovation and support, more questions emerge about the governance of the accompanying tools and technology. This article provides a summary of the FDA guidance on CDS, how CDS is defined, whether or not CDS is exempt from regulation, and how the FDA intends to enforce compliance. It also summarizes the FDA guidance on medical software, what software is exempt from regulation, and helps to answer some of the questions surrounding the digital health space.
One large healthcare system in the Pacific Northwest is moving machine learning technology from theory to practice. MultiCare Health System is using machine learning to develop a predictive model for reducing heart failure readmissions. Starting with 88 predictive variables applied to data from 69,000 heart failure patient encounters, the machine learning team has been able to quickly develop and refine a predictive model. The output from the model has guided resource allocation efforts and pre-discharge decision making to significantly improve patient care management activities. And the data has engendered trust among clinicians who rely on it the most for clinical decision making. This inside look at the application of advanced technology offers lessons for any healthcare system planning to ramp up its machine learning and predictive analytics efforts.
Care management is a tool for population health that focuses scarce healthcare resources on the sickest patients. Care management leaders need to know who those sickest patients are (or may be). The static risk models typically used for stratifying patients into risk categories only paint a partial picture of health and are ineffective for modern care management programs. Custom algorithms are now capable of predicting risk based on multiple risk models and multiple data sources. They help care management teams confidently stratify patient populations to paint a complete picture of care needs and efficiently deliver care to those who need it most. This article explains how custom algorithms work on static risk models to normalize risk scores and improve patient stratification, care management, and, ultimately, population health management.
With over 400,000 patient-harm related deaths annually and costs of more the $1 billion, health systems urgently need ways to improve patient safety. One promising safety solution is patient harm risk assessment tools that leverage machine learning. An effective patient safety surveillance tool has five core capabilities:
- Identifies risk: provides concurrent daily surveillance for all-cause harm events in a health system population.
- Stratifies patients at risk: places at-risk patients into risk categories (e.g., high, medium, and low risk).
- Shows modifiable risk factors: by understanding patient risk factors that can be modified, clinicians know where to intervene to prevent harm.
- Shows impactability: helps clinicians identify high-risk patients and prioritize treatment by patients who are most likely to benefit from preventive care.
- Makes risk prediction accessible: integrates risk prediction into workflow tools for immediate access.
Uncompensated care can cost large health systems billions of dollars annually, making outstanding balances one of their biggest costs. Propensity-to-pay tools help organizations target unpaid accounts by using artificial intelligence (AI) to leverage external and internal financial and socioeconomic data and identify the likelihood that patients in a population will pay their balances (propensity to pay). With propensity-to-pay insight, financial teams can focus their efforts on patients most likely to pay, and connect patients who are unable to pay with charity care or government assistance. Both health systems and patients benefit, as patients can avoid bad debt and organizations receive compensation for care they’ve delivered.
Those in Big Data and Healthcare Analytics circles will seldom hear the phrase “less is more.” In a clinical setting however, there is an important lesson to learn in regards to the effective execution of predictive analytics. We should not confuse more data with more insight. More data is simply more—as in more tables, more lists, more replicates, more clinics, more controls, more rows, tables of tables and lists of lists, etc. You get the idea. In short, for predictive analytics to be effective in a clinical venue, a specific focus will always trump global utility.
The opioid-related death rate in the U.S. has quadrupled since 1999, making more effective ways to predict opioid misuse a healthcare priority. A new generation of machine learning-enabled risk assessment tools promises to deliver broader and more relevant insight into a patient’s risk. With more comprehensive insight (including comorbidities, other substance abuse, the amount of medication prescribed, and the duration of opioid use), clinicians can make informed decisions when prescribing opioids and reduce the risk that patients will misuse, abuse, or overuse the pain killers. Clinicians will also be able to identify which patients might benefit from alternatives to opioid pain management (non-pharmacologic, multi-modal therapies, or care management programs).
Under value-based healthcare and the 2012 Hospital Readmission Reduction Program, healthcare organizations are more motivated than ever to reduce their incidence of preventable readmissions. Health systems can reduce risk of hospital readmissions by developing readmission risk scores tailored specifically to their populations. A risk model that meets the following five requirements will have significant predictive value and is most likely to achieve systemwide adoption:
- Identifies at-risk patients early.
- Separates patients relevant to the disease-specific identification method and intervention strategy from all other in-hospital patients.
- Uses organization-specific data to train a disease-specific model.
- Exceeds performance of existing models.
- Is developed in collaboration with domain experts.