Table 4.

Attributes used to predict whether the most frequently queried disease is the one afflicting the user.

Index	Attribute
1	The number of times queried for the most common diseases.
2	The number of times queried for the second most common diseases.
3	The ratio between the above two.
4	The fraction of users who asked about the most common disease queried by the user.
5	The number of diseases that were asked about more than once.
6	The number of diseases that were asked about more than five times.
7	The number of diseases that were asked about more than 10 times.
8	Indication of queries for drugs related to the most common disease queried by the user: Let M be a matrix of co-occurring diseases and drugs, where Mi,j is the number of users who asked about disease i and also about drug j, and let D be a matrix of queries for each disease, where Dij is whether user i asked about drug j, we compute an indicator of whether the user asked about drugs related to the most common disease they asked about, by multiplying the row vector of D for each user by the matrix M.