CS3600 Project 3 Solution

Figure 1: Example Bayesian network for medical diagnosis.
Source: http://song.bayesian.net/index.php/Bayesian_net
Probabilistic inference over Bayesian networks is a standard AI technique for medical diagnosis. Bayesian networks represent complex causal relationships between patient information, medical conditions, and symptoms. Probabilistic inference allows us to compute diagnostic queries, determining the likelihood of medical conditions given observed symptoms as evidence.Use the example Bayes net above as a prompt for the following questions.
Question 1
Recall that the na¨ıve Bayes assumption is that no effects of a cause are also causes of each other. If two effects are correlated it is because they are related to the same, underlying cause. The na¨ıve Bayes model provides an alternative representation for diagnostic inference. Draw a Bayes net representing the na¨ıve Bayes model for diagnosing Flu given its symptoms (assume the symptoms of Flu are every successor of Flu in the Bayes net in Figure 1). Which model (the Bayes net in Figure 1 or the na¨ıve Bayes model that you’ve constructed) is a richer representation? That is to say, is there anything we can represent with one model that we cannot represent with the other model? Answer:

The Bayes net in Figure 1 is a richer representation compared to the naive bayes model that we constructed because it can account for the fact that some effects of courses are also causes of each other. This is because the naive model assumes independence of factors. For example fever affecting Arthralgia is not accounted for in the naive model.
Question 2
SICKt−1 P(SICKt = T|SICKt−1) P(SICKt = F|SICKt−1)
T 0.7 0.3
F 0.5 0.5
Table 1: Transition Probabilities

Figure 2: First Order Markov Dynamic Bayes Net
Answer:

Question 3
Medical diagnosis with Bayesian networks are currently used as a decision support systems by healthcare professionals. An expert can input patient information and observed symptoms, and the decision support system outputs a set of possible diagnoses with associated likelihoods, but the final diagnosis decision is up to the medical professional. Why should we require a human supervisor to accept or override the decision of the AI diagnosis system? Name two (2) potential sources of error or unaccounted for situations for these Bayes net diagnosis models that are mitigated by having a trained healthcare professional make the final diagnosis decision.
Answer: Two potential sources of error for bayes net diagnosis models are the lack of sufficient data/medical knowledge and simple overestimation. There can be a case in which the disease is rarer and has less data collected about it. In this case, the doctor’s knowledge of the scenario would simply be more accurate. More generally, websites that seek to diagnose such as WebMD are joked about given the conclusions they sometimes arrive at given a set of symptoms. Human intervention would help mitigate misdiagnoses and make sure the proper treatment is suggested.
Question 4
Publicly accessible online services often use databases and symptom matching to inform users of possible medical conditions given a list of symptoms. These services do not provide diagnosis likelihoods. Could providing a free online service with Bayes-net-based medical diagnosis have negative impacts on human behavior? Could they have positive impacts? If you answered yes to either question, give one example. If you answered no, explain why not.