Exploring data helps us understand our world. Why something happened, how things are trending, and what constitutes a meaningful change.
Now, we have a vast set of tools in our data analysis toolbox. We can not only examine what happened in the past, but we can also predict what might happen in the future using predictive analytics.
Predictive analytics is a powerful tool that can help us accelerate the path to healthcare value, ultimately reducing healthcare costs while improving patient care.
Read on for an introduction to predictive analytics in healthcare, including the uses, benefits, value, and potential future of predictive analytics.
In this post we will cover:
- What Is Predictive Analytics in Healthcare?
- What Is the Role of Predictive Analytics?
- What Are Use Cases of Predictive Analytics in Healthcare?
- What Is the Value of Predictive Analytics?
- What Is the Future of Predictive Analytics?
What Is Predictive Analytics in Healthcare?
Predictive analytics answers the question: what is likely to happen next?
Using modeling and forecasting techniques, predictive analytics helps determine what has a high probability of occurring in the future. Physicians, researchers, medical specialty societies, pharmaceutical companies, and every other healthcare stakeholder can then use those predictions to provide the best possible care for individual patients.
To provide the most robust predictions possible, predictive analytics uses a variety of modeling techniques, from traditional hierarchical linear models to advanced artificial intelligence (AI) and machine learning algorithms. AI is a collection of technologies that can think and adapt on their own, while machine learning, which is a subset of AI, involves generating models to describe data with greater and greater precision as more data is introduced. [1,2]
In healthcare, predictive analytics leverages AI and machine learning to analyze historical data and forecast outputs such as disease risk for individual patients. [3]
What Is the Role of Predictive Analytics?
Predictive analytics is quickly becoming a cornerstone of personalized healthcare.
Using AI and machine learning, predictive models can intake huge amounts of diverse data for a single patient and forecast a patient’s response to certain treatments or devices, their risk of developing a specific disease, and their prognosis for a given condition.
What is personalized healthcare? This customized way of treating a patient is rooted in that individual’s specific medical history, environment, social risk factors, genetics, and unique biochemistry, among other characteristics. The key to personalized healthcare is treating a patient based on their specific attributes, instead of relying on population averages that aren’t one-size-fits-all. [4]
Predictive analytics is helping the healthcare system shift from treating a patient as an average to treating a patient as an individual, which can only improve patient care overall in terms of quality, efficiency, cost, and patient satisfaction.
Use Cases of Predictive Analytics in Healthcare
Predictive analytics is useful at every step in a patient’s journey, including diagnosis, prognosis, and treatment. Predictive analytics can also inform remote patient monitoring and reduce adverse events. On a more macro level, predictive analytics can improve care quality while reducing costs.
Leveraging predictive analytics can help answer questions like:
- What is the best course of treatment for this patient?
- Is this patient likely to experience an adverse event following a given procedure?
- What is the likelihood that this patient has a malignant tumor?
Here are a few key examples of predictive analytics in healthcare being used at various points in the patient journey:
- Diagnosis: Predictive analytics have been used to predict malignant mesothelioma diagnoses in a patient cohort. Patients diagnosed early can start treatment immediately, which improves their chances of survival—thus making prediction a critical tool. [5]
- Prognosis: Researchers used predictive analytics on physiological data from patients with congestive heart failure (CHF) to predict which patients were at greatest risk of readmission following a hospital stay. Using that information, physicians could implement interventions early to prevent the predicted readmissions. [6]
- Treatment: Clinicians have used machine learning-based predictive analytic models to determine the most effective course of treatment for chronic pain patients. [7]
Most importantly, predictive analytics can provide real-time clinical decision support at the point of care, making personalized healthcare as efficient as possible.
This is just a small sliver of the research being performed using predictive analytics in healthcare. As technology and analytic models advance, more opportunities to improve patient care using forecasting techniques will surely be discovered.
What Is the Value of Predictive Analytics?
Wherever there is data, predictive analytics can produce tremendous value. Leading organizations leverage predictive analytics to generate real-world results. Let’s look at an example below.
Michigan Bariatric Surgery Collaborative (MBSC) Uses a Predictive Calculator to Forecast Bariatric Surgery Outcomes
MBSC, a quality improvement collaborative funded by Blue Cross Blue Shield, leverages a powerful patient registry to improve bariatric surgery in Michigan.
MBSC uses clinical and patient-reported data combined with powerful predictive analytics to support clinical decision-making. Participating physicians use the MBSC registry and patient engagement tools to collect comprehensive preoperative data for potential bariatric surgery patients.
This data is then used to feed the MBSC Predictive Outcomes Calculator, a publicly available tool physicians can use to predict a patient’s weight loss, comorbidity resolution, and complication rate after bariatric surgery.
Specifically, the tool uses patient information on demographics, comorbidities, and other risk factors to forecast:
- Predicts weight loss at years 1, 2, and 3 for each of the available procedures.
- Predicts the likelihood of resolving weight-related comorbidities such as diabetes or sleep apnea.
- Predicts the likelihood of adverse events, serious complications and mortality.
Notably, the predicted rates for weight loss, comorbidity resolution, and potential complications are patient-specific, using risk-adjusted, real-world outcomes data from similar patients. These tools and other quality improvement initiatives helped MBSC and its members decrease rates of venous thromboembolism (VTE) by 43% and decrease post-surgical death rates by 67%.
What Is the Future of Predictive Analytics?
As we amass more and more data, predictive analytics will become both more common and more accurate.
Currently, many predictive models use traditional statistical methods, such as logistic regression, which are useful and can provide insightful results. However, AI and machine learning methods such as random forests, when implemented appropriately, can provide more accurate predictions. Ultimately, as more feature-rich data is collected and as the collection process itself improves, predictive analytics can take advantage of deep learning algorithms that can make better use of large and complex data sets. [8]
For example, deep learning algorithms can be used for image recognition, whether that be images gleaned from MRIs or other types of imaging technology, to automatically detect certain features. In contrast, a machine learning-based approach would require the radiologist to extract all the features from the image first. Deep learning simplifies the process by detecting all the features automatically, without requiring extra work by the radiologist.
Put simply, big data and predictive analytics in healthcare go hand in hand—more data means better predictions.
Additionally, the field of predictive analytics will advance to overcome its limitations. One current drawback is that predictive analytics is not able to provide insight into what might happen after an intervention or other change, which can be frustrating for researchers and clinicians who want to understand how patients might fare after a new treatment or as a result of a new hospital procedure. We expect predictive analytics will advance past this challenge, allowing researchers to forecast the future in a much more expansive and holistic way.
Predictive analytics will also be able to benefit from the massive leaps in computer processing power that make chewing through complicated algorithms possible.
Someday soon, predicting the future with accuracy under a variety of conditions will be the norm, not a fortune-teller’s trick.