Prescriptive analytics holds the promise to change the healthcare business immensely, but in order to deliver on its promise, healthcare systems need to take a different approach to the type of data that is factored into the advanced analytics equation. A more holistic approach to data will empower organizations to develop best practices for patient care and health IT administration.
Currently, data analytics are being used across most healthcare systems in the US on some level, but the sophistication of the technology used to find value in that data has a long way to go. There’s no doubt that collecting and analyzing a patient’s healthcare history data and information available directly through health and insurance records is valuable. The shortcoming is that the industry is too often stopping there, when the potential to create individualized care and analysis is dependent on a more holistic view of that patient’s life beyond what’s found in an electronic medical record.
The next phase of advanced analytics and individualized healthcare is to look beyond the data in front of us, and incorporate peripheral data to more accurately portray an individual. Healthcare systems have for years created data models and predictions based on generalizations about a certain group of patients – factors such as geographic region, age, sex and occupation. Grouping these populations and categorizing these risks can at times serve as a good starting point, but it doesn’t take into account a deeper understanding of the individual. The only way to get a truly accurate score for an individual patient within a population is to take peripheral factors into consideration when constructing a model.
This peripheral data could include information about the patient such as income data, family structure, census data, where they live and where they travel. Using outside data combined with pertinent health data will ultimately increase the quality of care for the individual patient, allowing healthcare professionals to get a highly accurate score of how to best treat patients and head potential diseases or readmission rates. This new data strategy aims to achieve sufficient volume and variety as well as being close to real-time. An accurate prescriptive score requires taking the entire ecosystem of data into consideration to generate a model, which will help to narrow the window of risk to hospitals. Once the model is built, it is necessary to maintain that model by making sure it is as up to date as possible as an individual patient and data shifts within their ecosystem.
The value of these maintained models is that they enable doctors and care managers to use critical data and information to support clinical, financial and operational decisions, and put them on the path to successful patient and business outcomes. Data does and will continue to grow, while decision makers will be faced with making decisions in less time. The success of a program handling these vast amounts of data is to rely on some type of automated advanced analytics and workflow capabilities to aid data scientists or business analysts. Healthcare systems need to invest in technology to automate data processes where possible, including the sourcing of new data, storing the data, analyzing the data and then integrating the advanced analytics into systems used by doctors and care managers. Automated advanced analytics identifies the highest risk patients across multiple areas of care, prioritizes the data, then prescribe the next best action.
In addition to expanding the data used for scoring, models themselves will have to be revised to account for inevitable changes in data sets – both existing and new. Many models used by healthcare systems today are still static, meaning they are not automatically adaptive to the uniqueness of each provider’s population. This lack of organic change can result in a larger margin of error. This error rate can and will translate directly into larger unforeseen expenses for providers and potentially degrade the quality of care patients receive. Trying to individualize models for specific patient populations within a static environment will result in high service fees and repeatable human interaction with models. The solution is combining a robust and dynamic data strategy with automated advanced analytics, which can be deployed across a wide set of profiles and adjusted in real-time to extract key, highly accurate insights about each individual.
Automated advanced analytics eliminates the need for healthcare technology companies to staff teams of programmers and data scientists to integrate techniques like machine learning into new and existing products. Rather, those high value employees can focus on the outcomes of the data to highlight weaknesses and prescribe better business models. Through automated advanced analytics, the complex problems that healthcare systems face can be answered, and both care and financial risk can be minimized, freeing resources to focus on critical tasks and delivering better care.