There were two major findings. First, we identified several features of Internet usage that correlated with depression. In other words, we found a trend: in general, the more a participant’s score on the survey indicated depression, the more his or her Internet usage included these (rather technical-sounding) features — for instance, “p2p packets,” which indicate high levels of sharing files (like movies and music).

Our second major discovery was that there were patterns of Internet usage that were statistically high among participants with depressive symptoms compared with those without symptoms. That is, we found indicators: styles of Internet behavior that were signs of depressive people. For example, participants with depressive symptoms tended to engage in very high e-mail usage. This perhaps was to be expected: research by the psychologists Janet Morahan-Martin and Phyllis Schumacher has shown that frequent checking of e-mail may relate to high levels of anxiety, which itself correlates with depressive symptoms.

Another example: the Internet usage of depressive people tended to exhibit high “flow duration entropy” — which often occurs when there is frequent switching among Internet applications like e-mail, chat rooms and games.