Predicting the Future of Wounds with AI and Analytics

How Hospitals are Leveraging Predictive Analytics to Improve Clinical Outcomes

In this blog post, we provide an overview of analytics that have been deployed in wound care in the past and examine how predictive analytics is now being applied to improve clinical and operational processes. We’ll review:

• The types of analytics traditionally deployed in wound care
• What predictive analytics are now being deployed in wound care
• How healthcare organizations are leveraging predictive analytics to improve their clinical and operational results today.

While data and analytics are already playing a pivotal role in the wound care space, solutions to date have largely focused on descriptive analytics, those that describe what has happened in the past or what is happening right now. But that is changing. Innovative hospital and affiliated wound care clinics are now using predictive analytics‚ algorithms based on existing and historical data that predict with a high degree of accuracy what will happen in the future.

Leveraging predictive analytics is enabling wound care providers to improve clinical decision-making and business operations, which leads to better patient outcomes, reduced costs and higher reimbursement. Predictive analytics rely on machine learning algorithms, a form of artificial intelligence, to compare a current situation (either clinical or operational) to past situations and determine what is likely to happen next. Predictive analytics can help providers leverage an astounding amount of data that no human mind, no matter how brilliant, could sift through. The algorithms in these analytics can analyze millions of pieces of data in seconds to recommend a course of action that delivers specific and targeted insights providers may never have been able to arrive at on their own. These analytical algorithms can provide suggestions or guidance that lead clinicians and administrators to better results.

Traditional Wound Care Analytics: About The Past or Present

Traditionally, analytics used in wound care clinics have been “descriptive analytics.” They present data on what has happened in the past or what is happening right now. Descriptive analytics work to  remove some of the noise and show with more accuracy  what is happening or has happened historically. Here are a couple of examples of how descriptive analytics are being applied in wound care clinics today:

• Business Intelligence (BI): BI usually takes the form of reports or dashboards that show current or past status and trends. Often BI  reporting is used to show a history of past medical results for a patient or a high-level, facility-level display of performance.
• Tissue Imaging Solutions: These  solutions utilize computer vision and machine learning algorithms to automatically identify the characteristics of a wound, such as its size.

Though sophisticated, tissue imaging solutions are still descriptive in nature. Clinicians use them during a patient visit to objectively assess the wound in real-time so they can determine the best course of treatment.

Predictive Analytics Help Healthcare Organizations See into the Future

Predictive analytics are a new form of analytics that are being deployed in clinical, administrative, and back office settings. These analytics use machine learning to identify patterns in historical data and then apply those patterns to current data to predict what is going to happen in the future. They allow clinicians, schedulers and others to see into the future and make far more informed and data-driven decisions than ever before.

Types of Data Elements Being Collected to Create Predictive Analytic Algorithms

Predictive analytic solutions for wound care, like the ones being produced by Net Health, look at hundreds of parameters and millions of data points, to create a predictive model of the data. That model is then applied to current clinical or operational data to predict the future results. That predictive algorithm produces results (with backing explanation) in real-time. Some of the data points used specifically in predictive analytic algorithms for wound care include:

• Wound characteristics such as tissue composition, percentage of slough, eschar granulation in the wound, pain, drainage exudate types, tissue types, dermatitis, induration excoriation, wound size, and  body location.
• Individual patient parameters like history of certain cancers or other diseases, types of diabetes, muscle weakness, etc.
• Vital signs at the point of care like blood pressure, heart rate and weight
• Patient demographic information, such as their geographic location and age

Predictive Analytics Improve Care, Reduce Costs, and Increase Efficiency

Predictive analytics algorithms are being used in more and more ways in the wound care setting to improve clinical results and the efficiency of providers, including:

• Clinical solutions, such as analytics that give insight into the wound trajectory/healing time or the risk of amputation.
• Operational solutions, such as a prediction of whether a patient will miss their next visit and recommendations on how to optimize their schedule for better outcomes.

Applications of Predictive Analytics in Wound Care

Identify patients at risk for self-discharge or missed visits. Clinicians and administrative staff do not have time to identify socioeconomic factors that increase a patient’s risk for missing a scheduled appointment. When patients miss visits or self-discharge, it results in sub-optimal care, an empty treatment room and reduced revenue.

• Predictive analytics can analyze hundreds of factors (e.g., distance between the patient‚ home and the treatment center, time of day of the appointment, weather, etc.) and identify which patients might be at a higher risk for missed visits or self-discharge. In real-time, a notification will alert the staff that the patient requires additional education or encouragement so they can take action that will keep the patient care on track. The alert also presents an opportunity to schedule patients most at risk for missing a visit during the same time block so that the clinic can better balance its patient flow and improve efficiency.
• Predict more accurate healing times. Predictive analytics are available that predict healing times given parameters such as the current size and shape of the wound, tissue composition, and oxygen levels.  

These algorithms also create a clinician sandbox in which clinicians can evaluate what-if scenarios to see how the application of a procedure could affect the healing trajectory of a wound. These algorithms can, therefore, provide clinicians with insights that improve treatment decisions, shorten healing time, and improve clinical outcomes.

Predictive Analytics can Improve Case Management and Patient Engagement

The more access to information case managers have, the better. And while we aren’t able to see into the future, predictive analytics is the closest we can get to that. As we mentioned, predictive analytics empower wound care nurses to better predict healing outcomes. What this means is that they can now better educate the patient about how to improve outcomes and they can actually conduct interventions at the point of care to improve outcomes. Additionally, predictive analytics may be able to help relieve some of the burden of documentation for case managers and providers, as well as the following:

• Increase clinical outcomes and patient satisfaction. Regardless of how the Centers for Medicare & Medicaid Services (CMS) chooses to reimburse the industry in the future, better outcomes and happier patients have a direct and positive correlation with ROI.
• Identify areas of improvement. The ability to compare actual and predictive results with internal and external sources of data creates opportunities to identify areas of improvement and successes to build on. And as these data sets become available through cloud-based systems and direct integrations, the results will magnify.
• Ease the burden of documentation and imaging. One of the most tedious tasks and efficiency-robbers in wound care is the documentation and imaging of wounds. Currently, there are cutting-edge products that help to simplify and streamline this process. And while highly effective,  there are still manual tasks that must be completed and data to be entered with each image. In the future, we expect these tasks to enter the world of automation, freeing up time for clinicians, improving accuracy, and driving down costs.
• Leverage predictive analytics to engage patients. Sharing the analytics with patients is a way to encourage them to participate more in their self-care, including making sure they don’t miss appointments. Some of the ways this could be implemented include giving clinicians the tools to decrease self-discharge and increase patient participation by giving patients a clearer timeline for results; show patients things they can do or stop doing that will have a direct effect on treatment results; propose treatment interventions better suited for the individual patients. A shift from blanket-like approaches to highly specified treatment plans could greatly move the needle for patients

When patients and providers are on the same page about what they want to happen, what is expected to happen, and how they both can best achieve those goals, it’s a win-win. Predictive analytics may be the bridge that accomplishes this.

The Importance of Integrating Predictive Analytics into Clinical and Operational Workflows

While predictive analytics offer new possibilities to improve clinical and operational workflows, it is important to recognize that they also require changes to existing practices. Minimizing these changes is critical to ensuring their adoption and wide-spread use. Seamless integration of predictive analytics within EHR systems minimizes disruptions and, therefore, is essential to helping providers and patients realize the clinical, cost, and efficiency benefits that they offer.

By integrating predictive analytics into existing systems, we can minimize the need for clinicians to learn new systems, develop data sets from scratch, or interrupt existing processes. In addition, predictive analytics must be implemented in a way that they complement and don’t interfere with existing workflows. This can be accomplished by providing clinicians with on-demand access to the predictive algorithms to assist their decision making, while not taking decision-making power away from the clinician.

Future Trends: Increased Applications and Even More Prescriptive Analytics

Predictive analytics has brought new capabilities to the wound care setting, giving clinicians and administrators better insights that allow them to act to improve future results. While there are numerous predictive analytics algorithms already in use, we expect the applications of predictive analytics algorithms to expand as more data from improved EHRs becomes available and the number of users increases and drives more innovation. In addition to more predictive analytic solutions, additional types of analytics will begin to surface in wound care settings to provide even more assistance to clinicians and administrators, including:

• Diagnostic Analytics: Used to describe why something happened. In medicine, diagnostic analytics describes the use of data to provide a clinical diagnosis. An example of this would be classifying/diagnosing wound type and cause from an image.
• Prescriptive Analytics: Implies action and allows providers to unleash the information contained in a dataset to help inform their decision-making.