OpenIDEO is an open innovation platform. Join our global community to solve big challenges for social good. Sign Up / Login or Learn more


An intelligent reminder app to aid early dementia patients in completing daily tasks and to reduce dependence on caregivers.

Photo of Saksham Gandhi
5 9

Written by

Who is your idea designed for and how does it better support family caregivers as they care for a loved one with dementia?

Companion is an intelligent reminder app designed for people with early stage dementia and their families. By easing daily struggles of early stage dementia, it aims to allow patients to be self-sufficient for as long as possible, to remain at home and to help sustain their current lifestyle. Companion collects user information and makes intelligent predictions, aiding them in routine tasks such as paying phone bills and grocery shopping. This eases the responsibilities weighing on caregivers.

Companion Overview: 

Companion is a reminder app that increases the self-sufficiency of early-stage dementia patients through three types of reminders: time-critical reminders, location-based reminders and family-based reminders. It does so by gathering information from various sources and using that data to ensure its users complete critical tasks.

One of the major hurdles faced by early-stage dementia patients is the inability to remain financially independent, as found in our preliminary research. Part of the reason for this is that financial obligations (e.g. paying phone/utility bills) are time-critical and inflexible. Helping users manage this is a crucial first step in reducing their dependence on others.

Time-critical reminders help in reducing dependance. By parsing through a user’s email account, Companion makes use of notification emails (e.g. an e-phone bill) to gauge the user’s most critical upcoming tasks and gives them timely reminders accordingly. Additionally, it notes when the task has been completed. Time-critical reminders would be of great help for users to meet basic routine financial obligations, eliminating the need for caregivers to constantly oversee the finances of the dementia patients.

Another major problem for dementia patients is remembering to complete routine tasks, such as grocery shopping or returning books to libraries. Reminders for such tasks can be made more effectively by basing them off of location. This means using information about the users’ current location, anticipated location, and even movements from the previous week. Suppose the app had previously received notifications about an overdue book, and the user was leaving for work.  If the library were on the path to the user’s destination, Companion would immediately remind them about the overdue book. Offering reminders tailored to one’s unique individual routine means that the user would be more likely to complete the task on time.

Companion also aims to strengthen family bonds. The feature of family-based reminders would allow family members to set reminders for Companion users, which the users could then accept or reject. This could be useful to set reminders for family dinners or general gatherings. The user could also opt to give family members access to their reminder list. In this manner, they could keep a tab on the user, helping the user if the need arises, while allowing them to remain largely independent. It also adds another source of information for the app in order to ensure it can track important events for the user and remind them about them.

The three type of reminders are not necessarily mutually exclusive, rather, three general classifications that guide us on how to remind the user about them. The three methods of sourcing for information (through email, location API and family) can augment one another to more effectively keep track of the tasks awaiting the user’s attention. For instance, if a user receives an email notice to pick up his/her package at a post office, there is a deadline attached to that notification (time-critical reminder). However, the user is more likely to take action on that reminder if it is given when he/she is close to the post office. In this manner, reminders can be tailored using aspects of time-critical and location-based reminders to maximise the effectiveness of the reminder.

Companion responds to the rapid increase in the amount of personal data that can be accessed through smartphones and the internet. Companies increasingly send bills to customers through email instead of traditional paper notices. We now have access to precise information about the location of our loved ones through smartphone GPS technology. It is imperative that we use this currently untapped world of data to protect the most vulnerable among us. Companion aims to achieve that, helping those most in need, and keeping them and their families happier as dementia becomes a greater problem for our aging population. 

User Experience Map (includes mock up illustration of app) 

Situation: 66 Year old Grace, suffering from early-stage dementia, realizes she forgot to pay her bills on time once again and has been charged a late fee. 

Solution: Grace turns to the Companion app, to avoid forgetting critical tasks in the future. This allows her to remain independent, rather than requiring her loved ones to constantly check on her. 

App mock-up: In-app list of reminder categories desired by Grace including phone bills, library book return, family dinner plan and grocery shopping reminders 

App mock-up: Intelligent reminders predicting useful information to provide Grace at the given time and location 

App mock-up: Family members can use the app to add reminders for Grace from their phones, Grace can then choose to accept the reminder (will be added to her list) , or she can reject it. Her family will receive a notification about her choice 

App mock-up: Basic reminder list for Grace to check her upcoming tasks 

Supporting Articles:

  • To help families keep patients at home for as long as possible, McClendon advocates expanding support services and training for caregivers.

  • A 2005 report from the Alzheimer’s Association showed troubling trends in care at the end of life. In a sweeping review of the medical literature, the investigators found that 71 percent of nursing home residents with advanced dementia died within six months of admission, yet only 11 percent were referred to hospice care, which focuses on comfort rather than active treatment

  • 42% of people aged 65 and above currently own smartphones
  • 74% of people between the ages of 50 and 64 own smartphones, increasing the potential for usage of the app as the years pass

What early, lightweight experiment might you try out in your own community to find out if the idea will meet your expectations?

Initially, we will reach out to local caregivers of patients with early-stage dementia and divide them into two different groups. The control group will have the patients try out the app, while the other will not. Using the data and feedback we acquire from the two groups, we will keep refining the features of our app. Once we are confident in successfully augmenting caregivers’ efforts, our objective of increasing patient self-sufficiency will be satisfied.

What skills, input, or guidance from the OpenIDEO community would be most helpful in building out or refining your idea?

Using design thinking inputs from the OpenIDEO community, we will use the feedback in the design of our app so that dementia patients can interact with the app interface as effortlessly as possible. Additionally, we will need as much help as possible in reaching out to patients, acquiring data, and testing our app in later phases of development. Any good product used by masses requires the input of many people and users to become a truly refined tool.

How long has your idea existed?

  • 4 months - 1 year

This idea emerged from

  • A student collaboration

Tell us about your work experience:

The team is comprised of Georgia Tech students with backgrounds in computer science, engineering and design. The combined skills of the team include, entrepreneurship, machine learning, cloud computing, front-end design and embedded systems. Using these skills, the team also competed in hackathons.

How would you describe this idea while in an elevator with someone?

Companion is an intelligent reminder app to aid early-stage dementia patients in completing daily tasks, and to reduce dependence on caregivers. Using machine learning techniques, it predicts the user's needs and issues reminders accordingly, instead of passively waiting for their inputs. It also uses location information to decide when to issue reminders, tailoring them for each individual.

How does your idea demonstrate our Criteria of Accessibility?

Our solution, the Companion app, is entirely software-based. It is designed so the app can run entirely on the user's phone, requiring no additional services in the background to support it, minimizing cost. Its accessibility depends solely on the user's access to smartphones. Dementia mainly affects those aged 65 and above. 42% of this age group currently own smartphones ( This number is expected to rise rapidly, increasing accessibility of the app

How does your idea demonstrate or plan to demonstrate scalability?

The power of our app would grow as more people adopt it. The app is designed to run entirely on the user's smartphones, meaning as more users adopt it, they also provide the means to run the app. This allows us to scale rapidly. Instead of posing a challenge, this would actually provide us with more data on what features of the app benefited users the most. We could then tweak our machine learning models, release free updates to the app, and better serve the users.

How do you plan to measure the impact of your idea?

The impact of our idea would be measured in terms of scale and effectiveness. Scale would entail the number of users who download and regularly use the app. This would initially depend on our promotion efforts. Beyond that, we hope that its sheer utility would advertise the app for us. Effectiveness would mean how much the app helped those who downloaded the app. If user surveys report that the app decreased their dependence on caregivers, effectiveness would have been achieved.

What are your immediate next steps after the Challenge?

Firstly, we must finalize our app's architecture. We have completed the individual functions of the app, e.g. email parsing, text clustering. However, they must be put together to work cohesively. Beyond that, we need to find partners to work with, to gain initial traction for the app. We could work with caregiver organizations for this, or even government organizations dealing with dementia. Finally, once we have a sizable user base, we could work on refining the app based on user feedback.


Join the conversation:

Photo of Susan Jackewicz

Saksham, I like the concept of your idea, especially cueing reminders in terms of time, place, and family/social support system. Your statement "It is imperative that we use this currently untapped world of data to protect the most vulnerable among us." is so important. How do you envision your app having a different value proposition from all the other generally available digital reminder/notification tools that are already available?

Photo of Saksham Gandhi

Hi Susan, thank you, we truly hope our app can achieve that goal. To effectively achieve it, we've made our app, Companion, differ from others in various manners.

The most basic difference between our app and the others is the integration of various sources of data, instead of relying on just one or two sources. This is why Companion reads through emails, analyzes user's location data, and even allows family members to set reminders. Using these sources, it then decides what to remind the user about. Beyond that, Companion is more active in its decision making. Current apps rely on passive decision making, i.e. The user enters what to be reminded about, whereas Companion predicts what the user may find useful to be reminded about.

There are further extensions we would like to make to Companion to set it apart more, but for now we have focused on the most critical areas as the first stage of our project.

Photo of Susan Jackewicz

Saksham, thank you for your explanation. I see how Companion could “envelope” a person’s actions throughout the day. It could really make a difference, esp. for someone with mild cognitive impairment (MCI), a diagnosis that sometimes is given early on in disease development.

Photo of Joanna Spoth

Hi Saksham Gandhi - great updates! We like your user experience map. We'd like to learn more about the methods you mention in your comment. Does your team have the skills to source and effectively use the information from emails, location, and family members? Or is that something you'd have to look elsewhere for? Additionally, does your team have the skills to build out the machine learning component? Thanks for your participation!

Photo of Saksham Gandhi

Hi Joanna. Thanks for the encouragement and we're glad you liked our work! Most of what we discussed in the post is within our means at our current skill level.

For instance, one of the challenging aspects of building this app was accurately grouping emails (e.g. phone bill emails and library reminder emails should be separated into different groups so that accurate reminders can be given to the user). Using machine learning clustering algorithms, we recently got that working. For certain other aspects, including getting information based on user location and family members, we have looked into it enough to know that we can achieve it.

The Companion app is definitely challenging to build, and there are certain areas where we may get face difficulties. For now, we don't intend to look elsewhere for the skills required. However, being students at Georgia Tech, we are well positioned to seek guidance from friends and faculty around us should the need arise.