The amount of patient data stored in Electronic Health Records (EHR) systems is vast, and continues to grow exponentially. One doctor can see over 2,000 patients in a year, and EHR manufacturers often have thousands of doctors with millions of patient records in their networks. The clinical knowledge stored in these medical records can have a huge impact on medical outcomes, and leveraging this data strategically can revolutionize the way healthcare is currently delivered.
Many EHR vendors foresee a future in which EHR software is imbued with artificial intelligence (AI) and machine learning capabilities, and connected to all sorts of new devices in the internet of things (IoT). The wealth of patient data stored in EHR systems lends itself to these new computer science applications. In a future not so far away, a patient’s health record would contain even more detailed data, from a much wider range of sources, such as remote sensing devices and wearable technology, capturing a patient’s heart rate in real time and feeding it directly into the EHR. With AI and machine learning algorithms built into the EHR, the software would be able to analyze a patient’s entire health record and allow doctors to take well-informed medical actions.
Several questions arise, however. How will this machine learning technology be integrated? Can doctors be sure that their networks are secure enough to ingest all these different types of data from multiple sources? Given the growth of clinical data, how can you be sure that your data is being properly and securely stored?
The potential benefits of fully integrating AI into EHR systems are very promising, and at the same time it is important to fully evaluate how the transition will take place, and what healthcare IT will look like as a result of the change.
What Kinds Of AI Are Already Part Of Health IT?
Clinical Decision Support (CDS) technology has become an increasingly essential tool in the delivery of healthcare, and CDS tools are often built into the EHR system. This technology was developed to assist doctors in analyzing vast amounts of patient data in order to determine the best course of treatment and to alert providers of any dangerous pharmaceutical interactions, or anything that a doctor might miss in the extensive detail of a patient’s health record.
A recent article from Health IT Analytics about CDS technology acknowledges the fact that “CDS tools are increasingly leveraging machine learning and artificial intelligence to power sophisticated analytics.” Algorithms for machine learning ingest large amounts of data, and output detailed results.
The use of machine learning technology in EHR software has been tested. In a recent study developed by Joon Lee, PhD, from the School of Public Health and Health Systems at the University of Waterloo in Ontario, in order to test machine learning algorithms on “personalized predictive models,” wherein index patients are compared to past patients in order to predict medical outcomes. The data scientists used “a variety of machine learning methodologies, including clustering, distance-based, and classification models, to develop guidelines for whether or not a data point is acceptable.”
Dr. Lee found that these algorithms were more and less reliable depending on the sample size: “Several of the models, especially the basic mortality probability model that simply counted the number of deaths among similar patients in the sample data, are only accurate within a certain range of the sample size.” In some of the machine learning methodologies, reliability dropped off when the sample size was greater than 12,000 patients, while others were “less susceptible to accuracy problems.” While the results were mixed, the researchers concluded that machine learning algorithms in EHR software can provide valuable insights into “risk stratification and the allocation of resources.”
Other studies have shown more promising results. A deep learning network demonstrated 100% accuracy in diagnosing breast cancer using 400 biopsy images. The tool was “consistently more accurate than human pathologists at identifying the delineations of tumors in whole biopsy slides.”
How Should Patient Data Be Governed?
Any discussion of big data analytics in healthcare will beg the question of data governance. This topic concerns data management according to federal regulations such as HIPAA, as well as the heathcare organization’s own standards of integrity. These efforts are intended to ensure confidence in the patient that their record is protected and confidential, and so that the data remains actionable to the healthcare provider when making decisions delivering patient care.
Other benefits of good data governance are the increased efficiency that comes from maintaining accurate patient data, and the accountability among the medical staff that these records provide. These benefits not only favor the patient, but the operation of the medical clinic as a whole.
Data governance becomes even more important in a scenario where you have patient data coming in through remote sensing devices, which are not located in the clinic, yet are interoperable with the EHR and feed data into it.
Interoperability Across Healthcare Systems and Devices
Interoperability is the way that different medical devices and EHR systems can communicate with each other. An interoperable system can collect data from a range of different sources and present this data to the user in an intelligible fashion. This brings the discussion closer to the IoT, where any electronic medical device can be connected to the internet, and therefore connect to other internet-enabled technologies, including EHR software.
Organizations like HIMSS have helped drive progress in the long-standing issue of interoperability by bringing different stakeholders in the industry together to collaborate and find solutions. There is still much work that remains to achieve interoperability, given the variety of different types of EHR software and different medical devices, which require universal standards and computer protocols, as well as new and innovative uses of existing medical technology.
What Is The Future Of EHR Technology?
The integration of AI and machine learning, and the potential of the IoT will be game-changing developments in healtcare IT, and push EHR closer to a recently conceived paradigm, Comprehensive Health Records (CHR). With the ability to collect these unimaginable quantities of patient data, and analyze that data algorithmically in the EHR, the provision of healthcare will take into account the total life experience of each patient, and that patient relative to the population as a whole.
With the great power of this knowledge, there is an increased need for security. This is necessary, not only to protect the patient’s privacy, but to ensure that all the data in the CHR is accurate, and the medical staff is accountable. If EHR manufacturers can collaborate and develop these technologies together, with sound data governance strategies and interoperable standards and protocols, the integration of AI and machine learning in an IoT-friendly environment, may not be so far off.