I haven't read the papers first-hand but did see the research presented and IIRC the point is that this doesn't just help the person to remember the facts of what they did (as your suggested approach would). Instead, it allowed them to retain the *actual memories* of what had happened. Having visual and audio records of the events of the day, and reviewing them periodically over the next couple of weeks actually helped these people retain their memories of the events. I'm pretty sure they said this extended to details not captured by the SenseCam, demonstrating that they weren't just remembering the material they'd reviewed.
So quite different from 'Did I remember to take my pill? Oh yes, here, I wrote it down...'
When I first heard about the SenseCam project it was the lifeblogging pitch and I thought it was cool but gimmicky. The research results they had for improving patients' memories really impressed me though.