Machine Learning Found as a New Solution to Cancer Treatment

Cancer treatment now seems possible as a new study has revealed that it can be cured with the use of machine learning. Regina Barzilay and her team at MIT has found out a new way via which even a person’s proneness to this life threatening disease can be detected. It has been considered as a miraculous finding and could prove to be a life saver for millions of people worldwide.

Regina had been researching for quite a long time along with her students at MIT and the team has recently achieved this breakthrough milestone. Further, she is currently working with a team of doctors in order to apply her super impressive cancer treatment findings on the cancer patients worldwide. Also noteworthy is the fact that this particular finding to cancer treatment could lead to totally revolutionize the cancer care landscape as till now, there has been no organic and trustworthy treatment available to this fatal disease.

Machine Learning Emerged As a Solution In Order To Take on To Cancer

She herself was diagnosed with breast cancer in 2014 and very quickly realised that finding good data related to this disease is extremely hard to find. And, according to her, without knowing the actual reason and the degree of its spread across your body, one can only rely on his doctor’s best guesses. Also when it comes to choosing the best drug and the treatment practice for the cure, an uncertainty and confusion prevails constantly during the whole treatment period. Besides, there is always a fear of its recurrence anytime.

Doctors Current Approach Towards Cancer Treatment and The Newly Found Effective Approach

Across various areas of cancer care, whether its diagnosis, treatment or prevention, the data protocol has always been similar worldwide. Doctors firstly try to map the patient’s information to obtain the structured data and then comes the need to run basic statistical analyses in order to find correlations. But still, the approach seems primitive and outdated compared to what can be achieved with the availability of latest tech tools powered with computer science advancements.

How Machine Learning Can Help

Barzilay’s research on natural Language processing (NLP) made it possible for machines to search and interpret textual documents like those which cancer patients receive as pathology reports. Moreover, applying NLP tools, she and her students were also able to extract clinical information from as many as 108,000 reports which they found out via the nearby area hospitals.

Further, the database which her team has created boasts off an accuracy rate of 98 percent. The next thing to which she is aiming for is to incorporate the treatment outcomes into it.

Also Read : Computers Being Trained To Predict The Outcome of Acute Diseases

But, Barzilay is also concerned about the highly complex and computational recommendations which could emerge out of application of these machine learning techniques and these recommendations might prove to be unexplainable.

Also under a joint research, along with Tommi Jaakkola(electrical engineering and computer science professor at MIT) and Tao Lei(graduate student), she is currently developing interpretable neural models that can justify and explain machine-based predictive reasoning.

Another study which she carried out with a researcher named Hughes made them capable of creating a database that can be used to monitor the development of atypias which can help identify the patients who are at risk of developing cancer in their near future.

According to her, machines are the best at making predictions and their use in procurement and analysis of cancer-patients related data could make it possible to completely cure this disgusting and fatal disease. She had strongly recommended to use all the available information related to cancer-affected patients to develop a model which could help suggest treatment measures and tactics.

Combination of Deep learning and Machine learning to Provide The Best Solution

Also along with Lehman and graduate student Nicholas Locascio, Barzilay is looking to apply deep learning algorithms in order to automate the analysis of mammogram data. Further, as the first step, the aim has been to compute the density and other scores derived by radiologists who analyze these images manually. The ultimate intended goal is to identify the patients that are more likely to develop a tumor before this tumor gets visible on a mammogram. Also, they are aiming to predict those patients who are heading towards recurrence after the initial treatment.

According to Barzilay, what she mean by ultimate success is, when we would be able to draw a lot from computer science and machine learning algorithms in a number of unexpected ways to develop treatments for many other such health-related diseases.