EML Blog #3: Machine Learning and The Elderly

In the next ten years, machine learning technology is going to improve in speed and accuracy. Sensors such as those in the increasingly popular wearable technology will become more ubiquitous, providing thousands of channels of data from which conclusions about the world can be drawn. These conclusions can be used to assist and protect the elderly who today often rely on others or end up taking risks just going about their everyday lives to maintain their independence.

Greater Independence

Many elderly people rely on their children, grandchildren, or professional assistants to go about their daily lives. Dementia disorders and the associated cognitive degradation often result in individuals losing the ability to do even basic tasks due to confusion and forgetfulness. An assistant can stand by them during their activity to help them remember what to do and how to do it, but this requirement could be removed with machine learning technology.

Wearable technology is just getting its start in 2016, and in the next ten years it will become ubiquitous. Eventually technology like Google Glass will become smart enough to detect the activity the wearer is performing and offer suggestions to remind the person what to do. This could include directions to destinations while walking or driving, reminders to perform tasks like watering the garden or feeding pets, or keeping track of things like grocery lists. These kinds of context-aware reminders and helpful hints will be useful to everyone, but for an old man or woman who has trouble remembering anything, a little voice in their ear reminding them where they left their medication could be a literal lifesaver.

Safety

Consider the smart glasses described above. A camera could watch the wearer chopping an onion and determine if he or she were doing so in an unsafe manner, warning them of potential injury. Perhaps they’ve forgotten to keep pot handles turned in towards the stove, where curious grandchildren can’t accidentally pull a bucket of boiling water down on their heads.

Fall detection is another useful application for machine learning. The “I’ve fallen, and I can’t get up” line originating from the 1980’s LifeCall commerical has become a meme in the last thirty years, but it is still a serious problem (so serious, in fact, that the successor to LifeCall, LifeAlert, has created a new, seriously disturbing, commercial to combat the “LOL old people falling, so funny” attitude). Sensors placed throughout the home would be able to detect a fall and call family members or paramedics, perhaps even assessing the severity of the fall before doing so. There is some optimization to be made here, as summoning an ambulance for a little scrape would be wasteful, but ending up stuck on the bathroom floor with a broken hip for two days waiting for help to arrive would be worse.

Medical Diagnostics

IBM’s Watson is already assisting doctors with diagnosing patients. Watson can read 40 million documents in 15 seconds, digesting an vast array of medical literature and use it to accurately determine the cause of patient issues. The faster problems are discovered and diagnosed, the faster patients can receive treatment. The elderly are particularly susceptible to disease, and so identifying illness in them quickly can help save their lives.

Sensors at home can assist doctors in catching health problems in the elderly people before those people even know they have a problem. Some problems, such as hemiplegia or Parkinson’s disease, can measurably affect a person’s gait. Classifier algorithms can analyze how someone walks and determine whether or not they should see their doctor.


Conclusion

To sum up, here are the five ways machine learning will assist the elderly in 2026:

1. Activity detection and guidance

2. Context-aware reminders and hints

3. Unsafe conditions detection and warning

4. Fall detection, weighing severity before alerting the authorities

5. Improved medical diagnosis in the home and doctor’s office


Sources

Sources are embedded in links in the text.

A single golf clap? Or a long standing ovation?

By clapping more or less, you can signal to us which stories really stand out.