korywat

[korywat]

Project Idea: Dementia Assistant

Dementia is a debilitating condition in which the patient, over time, eventually loses cognitive function. However, it is usually a slow decline of the person’s mental ability. This means that, for a time, it may be appropriate for the patient to have a caregiver, but not necessary for the individual to live in a hospital or receive care 24/7. This caregiver could either be a professional, a family member, or a client friend.

According to a research project from St. Luke’s hospital,  caregivers of patients with dementia deal with several stressors in their day-to-day work. The most significant of these stressors, by a large margin, is “Inability to manage stress levels due to lack of support from family or community and/or poor knowledge of the dementia condition.” This indicates that many caregivers are likely untrained about how to best handle their patients, and become stressed to the responsibility and necessary time placed into the subject’s care.

There are several important factors a caregiver must take into account during their work. For example, the caregiver needs to deal with “personality changes” in the patient, as well as instances in which the patient may not even recognize his family, according to alzheimers.net. The emotional toll, then, can be significant on the caregiver.

To solve this problem, I propose a robot that can assist these nonprofessional (and possible professional caregivers) in lowering their stress by taking away some of the responsibility and emotional requirements involved in caregiving for patients with dementia. In particular, this robot will be:

·     A routine-memory aid for the patient.

Part of the work a caregiver does is assist their patients in maintaining a normal schedule. That means eating, taking medicine, exercising, and doing other important activates that maintain the patient’s wellbeing. For example, according to caregiver.org, a caregiver may ask their patient what color of shirt they would prefer to wear during the morning.

Our robot can help assist the caregiver by asking these questions in the morning, and relaying important information to the caregiver. For example, the robot could let the patient know that he should take his medicine, and report back on whether or not it saw the patient do so to the caregivier.

·     Social interaction and instant help for the patient.

The robot would be able to interact socially with a dementia patient, able to converse in very light, basic conversation. This shouldn’t be a large issue for patients, since they will end up needing simple conversation as their mental capacity deteriorates. As well, the robot can be a portal for the patient to ask general questions to Google, care-specific questions that the robot can answer, as well as questions that need to be relayed to the caregiver directly.

This can free time for the caregiver, who may want to focus on other important parts of care, or otherwise just enjoy more free time. Further, the robot may be able to answer questions regarding who the dementia patient is/who the caregiver is as the patient begins to lose memory of her/her life. This can reduce emotional strain on the caregiver, as the caregiver will not need to remind the patient of who he/she is (and won’t need to deal with the emotional consequences of doing so as after as previously).

·     A monitor for patient behavior.

The robot should use advanced computer vision techniques to determine the mood of the patient, and can alert the caregiver if something is wrong or the patient is acting unruly. This way, not only will the patient be able to get help more quickly should something go wrong, but the caregiver can feel more comfortable leaving the patient alone for longer periods of time. This should relive caregiver stress, since he/she will not need to be around their patient all the time or otherwise constantly think about their care.

It is important to note that this robot is a good solution to this problem above other sources because it’s an effective interaction tool. The only viable competitor to this robot would be something like an automated personal assistant (ex. Google Home), since a voice-based interaction system is likely most ‘natural’ and easy to use for someone with deteriorating mental ability. This robot has several features that make it better than a standard, stationary voice assistant with extra caregiving features. First, we’re using the Kuri mobile social robot, which will be able to follow the patient and talk to him/her. As well, the robot has some facial movement/shows some emotion. These factors will allow the patient to feel a strong emotional connection (listed as important by many caregiving sources) with the robot and accept it as one of his/her caregivers, whereas it may be hard to form an emotional connection with a speaker (Google Home) with an assistant built in. One other major issue with the voice-assistant form-factor as it is today, is that voice assistants sit on a desk. The mobility of this robot allows it to effectively monitor the patient, as well as interact with him/her at all times via only speech (no carrying a phone or looking at a speaker to talk). The mobility also opens up the possibility to future iterations of the robot being able to help with more caregiving tasks (for example, a future version could get mobile manipulator functionality) without significant development effort to change existing software to take into account that the robot is no longer stationary.

Building any specific subset of functionality should be feasible given the time constraints and material constraints of the Capstone. Most of the above has already been accomplished using Machine Learning by large corporations. This is a large asset to us, since it should be feasible to hook into APIs (like the Microsoft emotion detector: https://azure.microsoft.com/en-us/services/cognitive-services/emotion/) to power a large portion of our “intelligent software.” As well, voice recognition software already exists that should be able to power our robot’s human-computer interface. Also, telecommunication tech already exists that can be hooked up to our robot (so that the patient can call their caregiver at any time and talk through the robot). Further, using the physical capabilities of the Kuri mobile social robot, we should be able to constantly “monitor” the dementia patient while at the same time always give them a reliable companion when they need help.

It is notable, however, that all of this functionality may not be feasible within 10 weeks; as such, we will likely need to pick and choose what to include and exclude based on group strengths and time taken to complete each piece.

Sources:

https://www.beckershospitalreview.com/quality/10-guiding-principles-for-patient-centered-care.html

http://www.nuhs.edu.sg/wbn/slot/u3691/Scholarly%20Activity/20150515_Laura%20Issues%20faced%20by%20Caregivers%20of%20Persibs%20w%20Dementia.pdf

https://www.caregiver.org/caregivers-guide-understanding-dementia-behaviors

https://www.alzheimers.net/caring-for-someone-with-dementia/

[korywat]

Prior Work: Caretaker Robots

Problem: Currently, human interaction is necessary for emotional support. People need this support to aid in “reaching goals, typically in health, education, and training concepts.” As well, older people and those who are sick need caretakers. This form of emotional support is very expensive and difficult to obtain, so it makes sense to create robots as a cheaper and more widely available option. Several human factors need to be accurately emulated for caretaker robots to be feasible: “speech, gesture, facial expressions, body language, and even physiologic data” need to be fully understood for a computer to be able to provide emotional support and effectively assist with personal care.

Solution: Maja Matarić, a professor at USC, co-founded company called “emodyed” to tackle these very problems with a team of skilled engineers. They are working on developing systems that can interpret large amounts of personal data. These systems will, in theory, be designed so that a computer can learn how to best provide personal care to each individual. Matarić notes that human behavior can be positively impacted with completely non-humanoid robots, as long as those robots have enough social affordances for effective interaction with users.” The companion robots that Embodyed aims to build will “revolutionize human-centric care and wellness by enhancing quality of life for individuals and families,” according to their website.

Sources:

http://www.embodied.me

https://www.acm.org/articles/people-of-acm/2017/maja-mataric