Assignment
Read the hypothetical case presented below and then respond to the questions. Submit your responses via file upload.
Hypothetical case of AlternHealth.com
The year is 2025, and following the COVID-19 pandemic several years ago, the trend toward telemedicine has exploded in the United States. Patients quickly figured out that they could be treated just as effectively over video chat as they could in a traditional brick and mortar doctor's office, and they prefer the convenience. Doctors also prefer to practice via telemedicine - the technology available has made it possible for them to diagnose and treat patients just as effectively over a remote connection as they ever could in-person, and this way they can see more patients without dealing with any of the added expenses of big, cumbersome offices. Because patients and doctors no longer need to be physically anywhere near each other, it was necessary to develop technology to help them find each other.
Professor Mehta and Professor Bedan anticipated this trend a long time ago, so in early 2021 they started a revolutionary new web-based company called AlternaHealth.com. This company uses advanced algorithms to match patients with the most appropriate doctor to treat their particular issue. The idea immediately took off, and within a year nearly every patient and doctor in the country began using the platform, and the clever Professors became unimaginably rich! The system was so efficient that all local, state and federal insurance plans, as well as all private insurance companies in the US, decided that all patients on their plans would be forced to use AlternaHealth.com if they needed medical treatment.
AlternaHealth.com matches patients to doctors through a highly sophisticated and top-secret algorithm. The exact way it makes its doctor-patient pairings is a trade secret, but we can tell you that it relies on powerful surveillance technology. It has access not only to all past medical records and personal identifying information such as a person's age, gender, ethnicity, home address, etc., but a trove of other information, such as social media posts, websites visited, tv shows watched, etc. It can access banking information and determine how much money a person makes and how likely they are to be able to afford out-of-pocket co-pays, what kinds of foods they eat, and how much alcohol or tobacco they consume. The system can also access cell phone and smart-watch data and understand how much a person exercises, who they spend time with, how much sleep they get, even whether they have harmful addictions that could impact their life expectancy.
Using all this information, AlternaHealth.com runs its top-secret algorithm and assigns all patients a confidential "patient score". To ensure accuracy, these scores are constantly normalized by comparing patients' data points to the data points of other people with similar data points. This approach ensures that people who have the highest patient scores are people who, in the past, have had the best health outcomes on average. Similarly, the patients who have the lowest patient scores share data traits with people who traditionally have more health problems and shorter life expectancy.
The algorithm evaluates physicians with a similar process, and assigns each doctor a score based on the outcomes for every patient the doctor has ever treated. These scores are similarly normalized against peers to ensure that the best doctors - ie, the ones whose patients tend to have the best outcomes and live the longest - have the highest "physician scores".
At AlternaHealth.com, we want to incentivize America to be as healthy as possible, so our policy is to always match the patients with the highest patient score to the doctors who have the highest physician scores. This works well because it encourages patients to improve their health score by adopting healthy habits. It also works for the doctors because it encourages them to work hard to provide the best treatment possible, so that they can be matched to patients with higher patient scores, who are better able to pay for their treatment.
In order to better motivate doctors, the algorithm also assigns compensation based on the health outcomes of their patients. Doctors whose patients have the best outcomes improve their doctor score and get paid the most money, while the lowest scoring doctors get less money. We consider it a win-win, and we know that the algorithm works well, because after four years the scoring system has remained remarkably stable - in other words, low scoring patients and doctors have tended to remain low scoring, and high scoring patients and doctors have remained high scoring. We also know that the system is fair, because it is entirely automated. No humans are involved in the scoring, pairing or payment calculations at all, so we know that human biases are not a factor.
Task
Using the principles we have discussed in this course,
A. What issues do you see with the methodology used by AlternaHealth?
B. What types of data could be generated by this process?
C. What legal and ethical considerations should be relevant here?
D. In addition to legal and ethical considerations, what issues do you see from a pure data analysis perspective?
E. Is the conclusion that the algorithm works well a sound one? Why or why not?