Joshua Waring

ITS 360 (001)

Dr. Shin-Ping Tucker

Homework 8, Due 4/5/19

 

What Would You Do?

1)     I would ask him how much the ai has been tested, and how frequently it is used for investing in the real world.† I would also ask him about the security of the software, and whether any potential issues caused by bugs in the software that could cause loss of funds on my part are considered the financial firmís liability.

2)     I would select the non-autonomous car, as I love driving to relax and unwind.† Also, I like the sense of control that driving gives a person.† That being said, a fully autonomous car might be safer and insurance for it might be cheaper, so I would also have to take that into consideration.† I think that a non-autonomous car would definitely win out in the end for me, especially since its cheaper, unless I had a really long and tedious commute that could be made more bearable and productive by using an autonomous vehicle.

 

Case Studies

1.1) †I believe MD Anderson should not accept this level of accuracy.† I think that when dealing with serious/life-threatening illnesses, the program should be held to a higher than 90% accuracy rate.† I also think the process the machine learning uses to arrive at its conclusion needs to be easier to analyze, and the process should be repeatable with the same results.

1.2) †MD Anderson learned the importance of having compatible data for the machine learning program.† They also learned that it can be difficult to impliment AI into health-related fields, as there are so many unique and variable cases.† MD Anderson also learned that there should have been an easier way to see the steps Watson used to arrive at its conclusions to help doctors analyze its results.

1.3) MD Anderson should ensure the data is current and not outdated.† I think they should ensure ahead of time that the process will be able to be converted into an quasi-easy to understand machine learning system that provides accurate and repeatable results.† I think they should have foreseen that choosing a new health-records system could potentially lead to having to restart the project.

2.1) Sophia DDM is a better solution, as it is easily acessable.† DDM also collects, corraborates and compares data for better, more accurate results.† DDM is also able to be used on a pay-per-case (SAS) basis, allowing smaller clinics equal accessability to it. I think DDM has a greater potential for success in the long run, as it has shown to be more useful than MD Andersonís attempt, and it is also easier for clinics and hospitals to use.

2.2) Patient confidentiality should be tied to HIPPA.† I think that personal information should be confidential in the sense that it not necessarily be tied to a specific identity, however I believe that it should be shared for the benefit of others, as it will provide better, more accurate results when using shared information.

2.3) Sophia DDM uses patented algorithms and machine learning to analyze and predict health issues in patients using shared health data.† It is successful due to the amount of data and genetic predictability, as well as its relatively low cost. †This makes Sophia DDM a good solution for both large and small hospitals due to its universal-ness and affordability. It seems that Sophia DDM will be a good way to analyze and diagnose health problems (ie, cancer), and hopefully it sees future development and improvement so that it can be applied to other health concerns besides just cancer.