How Machine Learning Will Transform Biomedicine
Humans live in an age of advanced technology. In the past few decades, we have advanced in many areas of our lives, whether it is transportation or communication. One of the most critical areas we have grown in is medicine and the pharmaceutical division. Technology profoundly influences the manner in which medical facilities work as well as the discovery of new cures and treatments. This article will explain how we can further advance in the medical field using machine learning models and generate data to make the diagnosis and curing of diseases easier for everyone.
One of the most significant aspects of machine learning in biomedicine is to have a robust machine learning model. A machine learning model is a file that is trained to recognize certain types of patterns [1]. The model is provided with a set of tested and verified data, and an algorithm that it can use to reason and learn from. Such machine learning models can be applied in various medical fields, such as the testing of vaccines. According to the Pharmaceutical Research and Manufacturers of America (2015), more than 800 medicines and vaccines are on trial to help treat cancer. It would take a long time to test all of these vaccines using current procedures [2]. With the help of a great machine learning model, we could simulate the results of all these vaccines. However, finding the data for this is time-consuming; considering that a lot of data is required and at times it holds the line between life and death. A way to counter this problem is to make finding data automatic too. This would entail collecting data from every possible field, whether from hospitals or a driver's licence. If scientists were to concentrate on this, it would make finding data enormously easier, and even though results may not be attained right away, this would be excellent for the future.
However, it is essential to know that data is never the end goal. As Professor Daphne Koller, the CEO of Incintro (a company specializing in the intersection of biomedicine and machine learning), said in an interview with Lex Fridman:
“Data is not the end-goal, it is the means [3]. The end goal is helping address challenges in human health, and the way to do that is to apply machine learning to build predictive models; machine learning can only really apply if it possesses sufficient data”
So although finding the data will be vital, the most critical milestone is making the machine learning models. Model building would be relatively easier than making entire datasets, mainly because such machine learning models are already in the works. According to a study referenced in Nature, a machine learning model was taught to diagnose certain diseases using a dataset, and it had about 80% accuracy [4]. Even though this might not seem like much, it is a massive advancement for humans, and using similar technology, we can get a lot further in the future with making machine learning models.
Once we have adequate data and exceptional machine learning models, it would be time to use them to get results related to biomedicine in all forms, whether it is diagnosing diseases or finding cures for them. This whole program would allow doctors to feed in data and receive solutions as an output. For example, let’s say that a doctor wants to find out whether a patient has brain cancer or not. Usually, they would have to spend a substantial amount of time doing an entire biopsy of the cancerous tissue in the brain of the patient, which would require a complicated analysis to determine if the tumour is present or not. However, with the availability of good data, all someone would have to do is enter the person's information into specific categories and the patient's issues would be presented immediately, whilst cross-checking with hundreds of thousands of people. This way, everyone would now have a way to examine their symptoms and check to see if they have any diseases, even without the help of doctors, entirely because of advanced machine learning models and data. On top of this, the cost of distributing this new technology would be zero as it is just a program that anyone with a device could access. This might seem too good to be true, and in a way it is. There is a possibility that the machine could wrongly predict illnesses, resulting in a disaster for the patient, yet, the chances would be lower than those presented by human error, making machines more reliable and the way to go [5].
Incorporating machine learning with biomedicine may seem like the best thing we can do today for the biomedical field, but will it really change anything? To check this, we can look through the Sustainable Development Goals (SDGs) of humanity and the first and most obvious goal this would tackle is Goal 3: ‘Good Health and Well-being.’ As explained before, machine learning will make better predictions on whether or not someone has a disease compared to humans, because humans can make errors as we have many things on our mind and judgement could be clouded. Another great SDG that machine learning can tackle is Goal 10: ‘Reduced Inequalities.’ Right now throughout the world, your health is heavily impacted by where you live. For example, in Canada you may have free healthcare and great doctors who can diagnose you and give you the help you need, but in many third-world countries this is not the case. There, healthcare is scarce and costs thousands of dollars, however, this can all change with the help of machine learning. Machine learning models are simply just a file on a device, anyone with a device will be able to use it, free of cost, reducing the inequalities that many face when it comes to health care. Finally, machine learning can go hand in hand with Goal 8: ‘Decent Work and Economic Growth.’ According to America’s Gross Domestic Product (GDP), healthcare accounts for 17.7% of the country's spending, which is 3.8 trillion dollars [6]. This could be reduced with the help of machine learning models. Being able to tell if you have a disease by entering information onto a device will reduce the cost of diagnosing diseases from thousands to nothing. With this we will be able to redirect this money to help the world's economy grow exponentially.
In conclusion, with the help of machine learning, humans will transform the future of biomedicine for the better. It will take some time to make machine learning models that can find data, but once the data is produced and the machine learning models get designed, diagnosing and finding cures for all diseases will be made easier for everyone across the globe.
References
[1] Q. Radich and E. Cowley, “What is a machine learning model?,” May, 2021. [Online]. Available: https://docs.microsoft.com/en-us/windows/ai/windows-ml/what-is-a-machine-learning-model. [Accessed 23 October 2021].
[2] SAS. "Machine learning: What it is and why it matters," SAS, n.d.. [Online]. Available: https://www.sas.com/en_ca/insights/analytics/machine-learning.html. [Accessed 19 September 2021].
[3] L. Fridman, “Daphne Koller: Biomedicine and Machine Learning,” YouTube, May, 2020. [Online]. Available: https://www.youtube.com/watch?v=xlMTWfkQqbY. [Accessed 20 August 2021].
[4] D. J. Park, et al. “Development of machine learning model for diagnostic disease prediction based on laboratory tests.” Scientific Reports vol. 11, no. 7567, 2021. Available: https://doi.org/10.1038/s41598-021-87171-5.
[5] D. Leibowitz, “AI diagnoses disease better than your doctor, study finds,” Towards Data Science, October 2020. [Online]. Available: https://towardsdatascience.com/ai-diagnoses-disease-better-than-your-doctor-study-finds-a5cc0ffbf32. [Accessed 20 October 2021].
[6] The Centers for Medicare & Medicaid Services, “Historical,” CMS, n.d.. [Online]. Available: https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/NationalHealthExpendData/NationalHealthAccountsHistorical. [Accessed 3 October 2021].