Clinical Fellowship Chronicles: How Will Machine Learning Change Medicine?
This is the second installment of a 12-part series that follows our clinical fellow Tiffany D’souza throughout her journey with The DICE Group. This month, Tiffany delves into the world of machine learning and explores its potential in healthcare.
Machine Learning (ML) is an emerging technology that’s often misunderstood. In medicine, many people view it as either a silver bullet for today’s healthcare challenges or a dystopian creation taking over the world.
Most of these assumptions aren’t grounded in reality, and most people don’t understand how this technology works. The more I learn about machine learning, the more I recognize its complexity and see a need for cross-field education and collaboration.
Innovating Healthcare With Intelligent Systems
Machine learning is the idea that systems can learn from data, identify patterns and make decisions or predictions with minimal human intervention. It’s a form of artificial intelligence where computers learn to finish a task they’re not already programmed to complete.
These machines scour years of data in seconds, so they can spot trends humans might miss. I already see how useful this would be in medicine. Throughout the patient journey, machine learning has the potential to transform:
Diagnostics
By feeding models relevant clinical data, machines can diagnose diseases faster. It’s making huge strides in Radiology — since countless X-ray scans are available to train models — and is also picking up speed in fields like Ophthalmology, Pathology and Oncology.
Let’s explore how machine learning could diagnose a case of breast cancer. Systems could correlate a patient’s mammogram results with their medical history and predict their risk of developing the disease. If an anomaly appears, the model could pick up the change and predict cancer’s onset, giving doctors insight that might stop the disease in its tracks.
Prognosis
Machine learning could make huge advancements in prognostic medicine. As this technology evolves, I see it predicting disease progression with near-perfect accuracy.
Consider the breast cancer example. After making a diagnosis, models could predict metastasis, mortality rates and even comorbidities. This will revolutionize prognostic medicine and cancer care.
Treatment
Complicated disease processes, like breast cancer, have a variety of treatment options. Based on a patient’s prognosis, machine learning could help doctors pick the best treatment plan, predict a patient’s drug response and even recommend surgery.
Eventually, machine learning will intervene in all levels of medicine. I see these systems helping us make difficult decisions, predict disease development and recommend treatment plans for a range of conditions. But these models need quality medical data to learn, and accessing this information can be challenging.
The Big Dilemma Around Big Data
Electronic health record systems (EHRs) house oceans of medical information. But due to:
- The sensitive nature of clinical data
- A lack of interoperability between health systems
- Challenges extracting quality data, and
- Gaps in data from undocumented events that happen at home…
…collecting a complete data set is logistically complicated. Once we have the data, it needs to be anonymized and cleaned to ensure the system produces accurate results. It’s a huge lift and many organizations don’t have the resources for it.
And yet the success of machine learning in healthcare hinges on quality data. Luckily, technology is being developed to address these obstacles.
Doctors can also be part of the solution by producing detailed, informative patient notes. Patient encounters, disease progress details and discharge summaries provide the foundation for solid data sets that can be easily extracted. If clinicians can work together to source and organize patient information, I’m optimistic about the future of machine learning in medicine.
Marrying Medicine and Machines
The thought of machine learning can be overwhelming for many physicians. But the more I learn, the more I realize it’s a trend to be embraced, not feared.
To immerse myself in this technology, I’ve enrolled in linear algebra, calculus and probability as many machine learning ideas are based on these aspects of math. I also intend to learn the coding language Python. My mission is to be a translator between the technology and medicine fields, and educating myself in these areas will help me reach that goal.
Innovation doesn’t happen in a vacuum. By learning to capitalize on this technology, we can build a world where machines streamline hospital processes, speed up recovery and improve patient care.
Tiffany D’souza
Tiffany is completing a Healthcare Innovation Fellowship at The DICE Group before applying to her residency. She is also an online fitness coach and social media marketing consultant who loves baking cookies, solving crossword puzzles and exploring the potential of healthcare technology.