Applications of Machine Learning in Healthcare

Data Abundance

According to Wikipedia, the term information explosion refers to the rapid increase in the amount of published information or data and the effects of this abundance. A major cause for this information explosion is the digitization of the various industries. The healthcare industry is no exception to this phenomenon and has digitized various key components such as booking of appointments, billing, reports of diagnosis etc. In fact, it is estimated that the healthcare system generates approximately a trillion gigabytes every year and this number is doubling every two years! Vast amounts of data and complex problems, have led to a gradual acceptance of machine learning applications in the healthcare industry.

Machine Learning To The Rescue

Machine learning is making use of the data available to aid doctors in diagnosis, analysis to identify trends and patterns in the collected data, drug discovery among various other applications. A lot of research is ongoing into the use of machine learning for both the diagnosis and prognosis of cancer. Google has come up with a model to detect cancer which achieved 89% accuracy compared to 73% accuracy achieved by a human pathologist. Despite the fears of the general population regarding AI replacing physicians, AI and machine learning will become valuable tools that will help speed up and optimize a doctor’s job.

Healthcare Areas Improved by AI

  • Operations and Logistics Optimization
  • Drug Development
  • Diagnosis and Prognosis
  • Outbreaks Prediction
  • Clinical Trials and Research
Korbit Blog - Machine Learning in Healthcare Series
Korbit Blog – Machine Learning in Healthcare Series

Applications of Machine Learning in Healthcare

There are three main branches of machine learning: supervised, unsupervised and reinforcement learning. Each branch has its own advantages and disadvantages and can be applied to solve a wide set of problems.

Example of classification (under supervised learning). The figure shows the model's predictions of malignant (1) and benign (0) breast cancer based on the tumor size.
Example of classification (under supervised learning). The figure shows the model’s predictions of malignant (1) and benign (0) breast cancer based on the tumor size.

1. Supervised Learning

For supervised learning, the algorithms need labeled data and will output an optimized model that could be applied to new data in order to predict an outcome. Take for example predicting diabetes or breast cancer prognosis. There are many applicable examples of supervised learning in healthcare since there is a massive influx of data coming from many sources.

Example of clustering (under unsupervised learning). The figure shows  the model's grouping of different cancer types.
Example of clustering (under unsupervised learning). The figure shows the model’s grouping of different cancer types.

2. Unsupervised Learning

Unsupervised learning is used on complex data sets where there are too many variables for a human to deal with. The objective is to simplify the data by clustering it into manageable subsets. This type of analysis could help find answers to questions like “What type of recurring qualitative and quantitative features do we find in high-risk cancer patients”?

Example of a feedback loop (reinforcement learning) where the model is predicting the next steps in the chemotherapy process based on simulations of patient data using rewards.
Example of a feedback loop (reinforcement learning) where the model is predicting the next steps in the chemotherapy process based on simulations of patient data using rewards.

3. Reinforcement Learning

Lastly, reinforcement learning is an automated way to make effective decisions by formulating the problem using rewards. Applications for these types of algorithms could be used for problems that involve prolonged and sequential procedures. An example could be the optimization of chemotherapy treatments.

There are numerous applications for AI in healthcare and as technology evolves, adoption will flourish. Accenture estimated AI can bring savings in the healthcare industry of up to $150 billion in annual savings in the United States by 2026

Open Data Access

It’s important to be aware of the intrinsic biases that could influence the predictions of machine learning and AI tools as well as their impact in the healthcare system and their implications for the stakeholders. Access to anonymized healthcare data should be made public (within certain constraints) and companies and research organizations should have the right to analyze these data. By opening up the data floodgates, more companies will be able to run and test algorithms improving their overall efficacy. This could help boost innovation in unprecedented ways. A restricted and limited access to this valuable data could limit the types of companies involved in the process and their overall level of innovation.

Algorithm Transparency

In addition, there is an increasing need to have more transparency on the assumptions, biases and inner workings of algorithms that could eventually be used in hospitals. This transparency would allow stakeholders to have a better idea of the limitations of such algorithms and a better way to explain certain outcomes rather than blindly following the instructions of a black box.

Improving AI Literacy in Healthcare

The healthcare industry should help demystify and promote education around the inner workings of AI and machine learning with doctors and patients alike. The world is already dependent on AI but is just unaware of its use in everyday life. People all over the world let AI recommend their next movie choices, songs and even follow it’s GPS directions. The question is, how confident would you feel following an AI’s recommendation when a human life is on the line?

Machine Learning in Healthcare Series

In this series of articles we explore the use of machine learning in the healthcare industry. Important concepts and algorithms are covered with real machine learning applications including open data sets, open code (available on Github and Google Colab) used to run the analyses and the final results and insights. Learn the machine learning fundamentals and concepts in the “Learn AI with an AI Course” with Audrey Durand. Enroll for free here → 

Applications of Machine Learning in Healthcare: Audrey Durand explains how linear regression and classification are use for breast cancer detection.

Subscribe to the Korbit Newsletter

Stay up to date with news, blog posts, code examples and new courses. The newsletter goes out every month.


Upcoming Articles

  • What Should I Use: Linear Regression or Logistic Regression
  • Predicting Cancer Recurrence Outcome With Machine Learning (Including Code)
  • Predicting Cancer Recurrence Time With Machine Learning (Including Code)
  • How to Run Code on Google Colab
  • Linear Regression in Machine Learning Explained 
  • Classification in Machine Learning Explained 

More Articles for You

Linear Regression in Machine Learning Explained

What is Linear Regression? Linear regression is a statistical method that allows us to summarize and study the relationships between …

We’re Hiring: Machine Learning Engineer

Open Position: Machine Learning Engineer for EdTech Startup We are looking for an extremely motivated full-time machine learning engineer to …

We’re Hiring: Data Scientist/Curriculum Manager

Open Position: Data Scientist and Curriculum Manager for Education Startup We are looking for an extremely motivated full-time data scientist …

We’re Hiring: Back-End Engineer

We are looking for an extremely motivated full-time back-stack engineer to help us build the future of online education

Korbit Technologies Raises USD $2M To Transform Education

MONTREAL, Oct. 2, 2019 – Korbit Technologies, a Canadian education technology startup, last week closed their seed round of over $2 …

— Featured —