Listen : Audio version of this article
It is safe to say that there are too many manual processes in medicine. When in training, I write lab scores, diagnoses, and other graphic notes on paper. I always know this is an area where technology can help improve my workflow and hope it will also improve patient care. Since then, progress in electronic medical records has been extraordinary, but the information they provide is not much better than the old paper charts they replaced. If technology wants to improve care in the future, then the electronic information provided to doctors needs to be enhanced by analytical power and machine learning.
By using this sophisticated type of analysis, we can provide better information to doctors at the point of patient care. Having easy access to blood pressure and other vital signs when I see my patients routinely and hopefully. Imagine how much more useful it would be if I was also shown my patient’s risk of stroke, coronary artery disease, and kidney failure based on 50 recent blood pressure readings, laboratory test results, race, gender, family history, socioeconomic status, and the latest clinical trial data.
We need to advance more information to doctors so that they can make better decisions about patient diagnoses and treatment options while understanding the possible outcomes and costs for each. The value of machine learning in health care is its ability to process large datasets outside the scope of human capabilities, and then reliably transforms the analysis of the data into clinical insights that help doctors plan and provide care, which ultimately leads to better outcomes, costs lower than care, and increasing patient satisfaction.
Applied Machine Learning in Health Services:
Machine learning in medicine recently made headlines. Google has developed a machine learning algorithm to help identify cancer tumours on a mammogram. Stanford uses deep learning algorithms to identify skin cancer. A JAMA article recently reported the results of a deep machine learning algorithm capable of diagnosing diabetic retinopathy in retinal images. It is clear that machine learning places another arrow in the vibration of clinical decision making.
However, machine learning is suitable for some processes better than others. Algorithms can provide direct benefits to scientific disciplines with processes that can be reproduced or standardized. Also, those who have large drawing datasets, such as radiology, cardiology, and pathology, are strong candidates. Machine learning can be trained to see images, identify abnormalities, and point to areas that need attention, thereby increasing the accuracy of all these processes. Long-term, machine learning will be beneficial for family practitioners or internists at the bedside. Machine learning can offer objective opinions to improve efficiency, reliability and accuracy.
On Health Catalyst, we use proprietary platforms to analyze data and return it in real time to doctors to assist in clinical decision making. At the same time a doctor looks at the patient and includes symptoms, data, and test results into ESDM, there is a learning machine behind the scenes seeing everything about the patient, and encouraging the doctor with information that is useful for making a diagnosis, ordering a test, or suggesting preventive filtering. In the long run, the capability will reach all aspects of medicine when we get integrated data that is more useful. We will be able to combine larger data sets that can be analyzed and compared in real time to provide all types of information to providers and patients.
Ethics of Using Algorithms in Health:
It has been said before that the best machine learning tool in health care is the doctor’s brain. Could there be a tendency for doctors to view machine learning as an unwanted second opinion? At one point, automatic workers worry that robots will get rid of their jobs. Similarly, there may be doctors who fear that machine learning is the beginning of a process that can make them obsolete. But it’s an art of medicine that can never be replaced. Patients will always need a human touch, and a caring and caring relationship with the people who provide care. Both machine learning, as well as other future technologies in medicine, will eliminate this but will be a tool that doctors use to improve ongoing care.
Healthcare needs to move from thinking about machine learning as a futuristic concept to seeing it as a real-world tool that can be used today. If machine learning has a role in health care, then we must take an additional approach. We have to find special use cases where machine learning capabilities provide value from certain technological applications (eg, Google and Stanford). This will be a step-by-step path for incorporating more analytics, machine learning, and prediction algorithms into everyday clinical practice.
Medicine has a method for investigating and proving that treatment is safe and effective. This is a long trial process – and bases the decision on evidence. We need this same process in place when we look at machine learning to ensure its safety and efficacy. We need to understand the ethics involved in giving away some of what we do to a machine.
There are unlimited opportunities for machine learning in health care:
Some people might ask whether this is only a technological mode or whether it will provide true value in health care. Health Catalyst believes that the introduction and widespread use of machine learning in health care will be one of the most important life-saving technologies ever introduced. We believe that opportunities are actually not limited to technology to improve and accelerate clinical, workflow and financial results. The following are a few examples:
Reduce Readmissions. Machine learning can reduce re-acceptance in a way that is targeted, efficient, and patient-centred. Doctors can receive daily guidelines about which patients are most likely to be accepted again and how they can reduce that risk.
Prevent hospital-acquired infections (HAI). Health systems can reduce HAI, such as central line-related blood flow infections (CLABSI) —40 per cent of CLABSI patients die – by predicting which patients have a central channel that will develop CLABSI. Doctors can monitor high-risk patients and conduct interventions to reduce those risks by focusing on patient-specific risk factors.
Reducing the Duration of Staying in a Hospital (LOS). The health system can reduce LOS and increase other outcomes such as patient satisfaction by identifying patients who tend to experience an increase in LOS and then ensure that best practices are followed.
Predict chronic disease. Machine learning can help the hospital system identify patients with chronic diseases that are undiagnosed or misdiagnosed, predict the possibility that patients will develop chronic diseases, and present patient-specific preventive interventions.