The Conepts Of Artificial Intelligence & Machine Learning

Artificial Intelligence is a field of study where we teach computers how to learn, Machine Learning is one way of doing this, by designing code or programs that teach computers things over time. We have to select the game, formulate the goal in the language computer can understand and train it to optimize the score. This is how we teach cars how to drive, computers how to play the game of go and Alexa to understand the “Say your preference of coconut water”. The goal here in our case would be to suggest doctors the test that’s needed next in order to diagnose a disease.

Based on a decision tree and data driven strategy I believe this can be done. This would have lot of space for ethical issues, as one might be alright if an application such as Google maps or Yelp is tracking the location data, but that might not be the same when it comes to their Health records and data usage. Patients should be alright to share their test reports in order to train the data and test it against another person who is suffering. Data needs to be cleaned first, formatted in order to feed to the system and data drives the results. There is a possibility of the system giving incorrect or not valid treatment or test suggestions but that acts as an additional set of information to the doctor. In our case, we had couple of doctors whom we were contacting for second opinions. Imagine if there was this data driven analysis on hand and suggestions too. May be then we would have eliminated unnecessary biopsy’s or tests in the process of diagnosis.

There is a lot of work that is being put in already on Machine Learning in Healthcare by big companies such as IBM ( (Devinsky, 2016), Google and many other start-ups too. I have gone through this interesting project called TREWScore (Saria, 2015), where Machine Learning is used to predict Sepsis in a patient. They quote that it need not be in specific to Sepsis and that the system can be extended to detect other diseases using profile of symptoms such as Diabetes, Cancer, Parkinson’s etc. The challenge here would be fixing on one data model as the logic or decision tree varies from disease to disease based on the respective symptoms. It would also vary based on the population sets with respect to geographical locations. It could be a combination of over populated country with low cases and less populated country having more cases of a particular type.

Other factors to consider would be age, gender and health habits of the patient. So, do we start with one set that considers all the factors and change it according to the region, state and country still remains a question. It does not end there, who is going to pay for the new machinery stays as an open question, will the insurance over it, will the state government pay or will the load be on the individual pockets to obtain these services. Securing all the data will be a big problem in taking this path, but everything getting electronic these days, I believe putting them in some better use wouldn’t be a problem. Someone somewhere surely has your data, so why not share your story and scenario to help others who might need them.

Machine learning option enables to take decisions wisely with the help of some additional input from a trained smart system. It’s correctness assessment has lot of gaps to be fulfilled as health is not a simple math that a machine can perform. But we can definitely use the advancement in technology as a helping hand in making the decisions easier for the health experts. Being in 21st with science advancement, learning and machines providing a helping hand in many sectors, coming centuries might get better healthcare with what we do today. As we are growing, new diseases and challenges are coming up along with the advancements and so handling all these with time is mandatary to survive for long.

I believe we can start forming the data sets for a city for one disease and format it as we see more cases and then try to do that region wise. Hospitals, health centers, diagnostic centers, government, patients and families all together can make this work. I would be the first one to share my father-in-law’s case reports, without being emotionally locked up, if that can help in curing and treating any other individual suffering from a similar situation. I have previously worked on trying to analyze the diversity of citations depending on the location of journal published and on authors affiliation. I initially started working on a set of political sciences test data articles but later I have realized that my code works for any Harvard reference style paper (Leelu, 2018). That makes me strongly believe that at some point we can build a reliable system with minimal error.

End of the day we are just providing an additional source of information to human and helping him take decision by eliminating options. If I am given any chance and if I can get access to sample test data and by collaborating with a physician who has experience with such cases and knowledge, we could work on the programming. Yet again, I end up here saying I am an Information Technology person with no complete knowledge on the Healthcare and Health domain. This collaboration can do wonders I believe. With all the study so far, I have understood that any solution related to Health is not so easy to implement as it’s related to existence of life and about survival. That I have to think about a vast number of challenges and factors, I really look forward to working in this felid with my computer science knowledge and do my part for my family and for my fellow beings. Much respect to every person who is contributing in this field.