GCI-313: The Mingling of A.I and OpenMRS
How A.I can be a positive force in OpenMRS
As machine learning has grown to become more dominant and accessible in a multitude of fields, the healthcare sector arguably has the most potential for immense gains. Being so, the OpenMRS developer community has a great opportunity to implement many machine learning features that will not only improve the quality of diagnoses from doctors using OpenMRS, but also make it more user-friendly and universal.
Using Predictive Analysis
With the emergence of accessible and easy-to-use Python libraries like numpy, matplotlib, seas, and TensorFlow, it becomes more apparent that the next step for OpenMRS to develop its medical records is by using predictive analysis: the implementation of machine learning algorithms to determine correlations among variables. Predictive analysis would greatly benefit OpenMRS because rather than displaying basic information about a patient like allergies, ethnicity, and so on, machine learning algorithms could also inform doctors of potential risks the patients might face based on that data. Because of this, OpenMRS would have more directly meaningful data for doctors and could provide information that doctors can’t see by only analyzing one patient at a time. In fact, it wouldn’t be surprising if these algorithms found correlations between two variables that weren’t previously established due to how vast OpenMRS’s dataset is.
Even with all this data, this is not an impossible task either because of libraries like numpy which are specially designed to find relationships between large sets of data efficiently. There could be a new module in Python dedicated to implementing these algorithms and skilled developers with data science libraries in Python could create the frameworks needed to analyze the vast data in OpenMRS. The developers that normally work on the Javascript portion of OpenMRS could then display the data cleanly for doctors to see. For example, there could be a new “risks” category, which displays the highest percentages of risk of certain diseases in a patient. With data like this, doctors would not only save time and money having improved data but would also save lives!
Build better voice apps. Get more articles & interviews from voice technology experts at voicetechpodcast.com
Crossing Language Barriers
While natural language processing is still in its early development, it doesn’t change the fact that it is still useful enough to translate between languages and convert sounds into written text. Considering that many doctors don’t speak English and there are too many different languages doctors speak, it would be of great use to OpenMRS to create rough translations from one language to another using natural language processing algorithms. This would make OpenMRS more universal in the medical community as more doctors could use it and would also make it easier for doctors who may have trouble with English (for example) to use OpenMRS. Also, because miscellaneous files could be added to a patient’s record like videos, natural language processing algorithms could create subtitles for those videos, which would allow doctors easily understand what is being said in a video if the audio is unclear or if they are not fluent in the language the video is in. This, in turn, would make the miscellaneous section of OpenMRS more useable and more reliable.
The Power of Analyzed Images
It is not far-fetched to think that several machine learning algorithms could also edit images that may be associated with patients in ways that are meaningful and useful. For example, if there are radiology images of the brain, and somehow it was incomplete or a part of the image got cut off, machine learning algorithms could offer an alternative image of what is to be expected. Moreso than that, they could also edit the image in meaningful ways such as color-coding different parts that seem significant to the model based on its predictive analysis of other components of the patient’s record. Information that is analyzed in such a way would not only lead to new insights for doctors, especially if they have limited resources, but also would make OpenMRS more sophisticated and it would be one of the very few medical record systems that assists the doctor in doing their job. Doctors and machine learning models aren’t perfect, but together, they might just be!
The Future of OpenMRS
Based on where OpenMRS is right now, it really seems like predictive analysis would make it much better as it would be a system that actually analyzes data in useful ways for doctors. Natural language processing and image processing would be great add-ons as they would simplify the jobs of the doctor using an accessible tool, that being OpenMRS. Regardless, there are many different directions without how machine learning could be used in OpenMRS, and I think that is the most exciting part about its future!
