Mamita Health

An SMS-based messaging solution powered by GPT-3 and AI to provide answers to childbirth and pregnancy-related questions. 

Our solution is an SMS-based messaging system designed to educate mothers about maternal health through simple checkpoint messages throughout the pregnancy and on-demand question answering. The use of GPT-3 and NLP will ensure mothers are receiving the latest information specific to the conditions of their region. Natural Language Processing and Natural Language understanding will be utilized to make questions and answers clear and easy to understand (in both English and Spanish) for mothers in Bolivia. Through routine data collection from women (ex: asking about pains, morning sickness, eating habits, sleeping patterns, etc), our solution provides valuable insights based on patterns to help prevent complications before, during, and after birth. Through our referral program, we are able to identify potentially threatening situations and direct mothers to partner healthcare centers to seek medical attention. 

The specific problem and area that we intend on targeting is maternal mortality in Bolivia. Bolivia has the highest maternal mortality rate in the Western hemisphere (160 per 100,000 live births) and this is a direct result of poor maternal education. Many women in developing countries lack access to the internet and encounter accessibility issues when trying to get healthcare. High maternal mortality rates in many developing countries are a result of birth-related infections caused by a lack of caution and care during pregnancy. High maternal mortality rates are a combination of poor healthcare infrastructure and long travel distances coupled with low accessibility to quality healthcare. In 2019, the government of Bolivia built 2,710 clinics to increase accessibility yet still only 20% of the rural population has access to proper healthcare. The Unified Health System in Bolivia guarantees free healthcare, but the government is unable to make this goal achievable and sustainable. Doctors needed a budget of around $1 billion USD but only received about $200 million USD. Because of the lack of funds, there are not enough supplies or facilities available to provide the healthcare and education that so many Bolivians need. Even when citizens have a right to free healthcare, there are additional boundaries that may prevent them from getting the help they need. Bolivia’s rural areas tend to be much more burdened with poverty than urban areas. 

Our solution is specifically catered towards mothers in Bolivia. Our pilot plan is to implement in Lake Titicaca - specifically in small indigenous communities around Lake Titicaca which is situated near the city of La Paz. Despite communities lacking internet access, accessible healthcare will make it possible to successfully pilot our solution. There are approximately 14 health care centers situated in the region around the lake. In Bolivia, there is a lack of maternal education which makes it difficult for women to be aware of the best practices to take care of themselves and their babies. Specifically, in the indigenous regions around Lake Titicaca, there’s also a lack of maternal education resources. Ideas of maternal health are generally passed down through generations and institutional healthcare isn’t leveraged within indigenous communities. Our solution addresses all these problems. Through our partnership with MSI reproductive choices, we’ll be able to provide women in indeginous communities and remote areas with needed resources; our SMS solution will be able to effectively deliver maternal education. These women will be referred to healthcare centers in the neighbouring areas if necessary and will have access to quality answers to all of their pregnancy and healthcare questions. Best practices to take care of themselves and their baby will be implemented into their lives as well. 

We’ve taken many steps to get a better understanding of where we intend on piloting. We reached out to doctors and healthcare administrators to get a better understanding of the Bolivian healthcare system and their infrastructure. During our conversation, we ran our idea by them to to receive validation and if we would actually be able to implement in Lake Titicaca. The doctors approved of our idea and they said they would be willing to market this idea to women in the nearby area. We also talked to many Bolivian women to get a better understanding of their day to day lifestyle and seeing if they could actually use our SMS bot. We talked to families and tried to understand some of the sentiments regarding maternal healthcare and while we understand that maternal education is considered a taboo and difficult to talk about topic in the country, many organizations in Bolivia have succeeded at bringing awareness to issues concerning women’s health. We’ve also been able to reach many women in Bolivia through social media platforms like Facebook who’ve been able to give us more insight into the problems their communities face regarding maternal healthcare and validation around our solution. We have validation from the VP of MSI Reproductive Choices (which gives contraceptives to women in many developing countries), Andrew Seddon and Karima Ladhani (founder of Barakat Bundle).

The prototype stage is ideal for our solution. We’re currently working on establishing partnerships with many organizations and improving our prototypes based on feedback and validation we’ve received from a small group of women trying out our solution. We still need funds to build out our model since the costs associated with just making this service possible are high (detailed in the link below). Our goal for this year is to establish partnerships and secure funding to provide our solution to 5,000 women in Bolivia. The deck below details more in-depth on the steps required to get to the point of the pilot. Until then, we’re testing on a small group of users and working to improve the functionality of our solution (allowing referrals, connecting with potential partners, working on an effective data collection mechanism, etc.). 

Slide deck

GPT-3 is trained with huge internet datasets, it’s able to process and understand text effectively. The AI is constantly updating the database with new information to ensure the model stays relevant and is always up to date with the latest information. GPT-3 uses data compress while it’s training/consuming data. It converts the word input into a vector. After this is done, the model unpacks the compressed text into human-like sentences. Through this process of unpacking and packing data, the model’s accuracy is further developed. The second component of our solution is Natural Language Processing. Natural language processing would allow our solution to condense language and summarize text in meaningful ways for women in Bolivia in both Spanish and English. NLP is used for the processing of written text. After the text has been processed, the GPT-3 is used to retrieve data from a database. The system also has theory from semantics to guide the comprehension based on Bolivia’s literacy rate and English language education. Competing semantic theories of language have specific trade-offs in their suitability as the basis of computer-automated semantic interpretation. These range from naive semantics or stochastic semantic analysis to the use of pragmatics to derive meaning from context. Semantic parsers convert natural-language texts into formal meaning representations. These 2 technologies allow us to deliver effective and efficient maternal education to mothers. Please see more on our technology below (it's more in-depth and details everything further).

Slide deck

For the moment our solution isn’t in use by our target market. however, we have been testing the solution with a smaller group of women to receive feedback and iterate on our technology and models. By the end of the year, after launching a pilot in the region around Lake Titicaca in Bolivia, we plan to serve 5,000 women. We know this is a feasible target based on information from healthcare centers we reached out to and the understanding we've managed to achieve of the environment and status quo in the communities we plan to pilot in. While that number may seem small, we expect it to grow rapidly in the coming years as we implement more pilots in regions and continue to grow our network of partnerships and healthcare centers willing to refer their patients and support this mission.

Our impact goals consist of 2 objectives: increasing institutional healthcare visits and increasing the rate of maternal education. Our solution intends to increase the number of women who seek help by raising red flags earlier on so they can receive the right support. To measure this impact, we’ve connected with healthcare centers to help track our growth and help access more potential users for our solution. Currently, Bolivia lacks a structural system of providing maternal education to women. Our solution will fix this issue by prompting women to relay back information they’ve been given and identify the next steps for their actions after being given the information. This will help us not only improve our algorithms and models to provide more accurate and understandable answers, but also give us a good starting point of understanding the potential impact our solution is having on the lives of Bolivian women. Our plans to achieve this goal are to connect with more healthcare professionals and establish a firm initial user base. Through this user base, we intend on partnering with MSI reproductive choices. With this partnership, we’ll be able to outreach to more mothers in Bolivia and they’ll be aware of our service. Establishing advertisement partnerships is vital to making our solution sustainable.  

We went through more in detail about how we plan to achieve these goals in the slide deck

In order to measure the impact of our solution, we’re going to track the number of users. For our pilot, our goal is to start with 5000 users and we will track the increase/decrease of our users. Success for us is to notice a 15% increase in users over the months. Tracking this data would allow us to track conversion rates and the effectiveness of our solution. If our users are going up, our distribution channels are effectively reaching new users. With access to data such as recurring text messages, we can measure how successful our product is. If users are returning, they are most likely satisfied with the service. On a monthly basis, if at least 50% of the users are returning, we know that our current system is effective and working. Tracking message use would allow us to improve our technology to ensure that we’re meeting our goals. The last data point that we intend on tracking is the increased number of hospital visits. One of the biggest issues we identified was the lack of healthcare visits and we hope that solution prompts mothers to visit the hospital often. We’re going to be in contact with doctors in the place of our pilot and if we’re able to see an increase in the number of visits, our solution has been a success. Seeking medical attention is vital to lowering the maternal mortality rate. 

One of the biggest barriers that we’re currently facing is trying to cover the finances for the first year of iterating on our service and initial prototype and pilot in Bolivia. According to our financial breakdown (more can be found in our slide deck), we’re going to be in debt of about $5,500. If we face any complications, our debts might increase, but we intend on making this money back through our ad revenue which we will secure during our second year. As high school students, it's challenging for us to make up for the costs of making this type of service free for every woman. Despite this, we're currently negotiating with Twilio to make SMS messages cheaper for us to send so we can potentially decrease the costs we will be spending on SMS messaging. In terms of market barriers, we have strong connections to doctors and organizations within Bolivia so establishing a market and reaching women who would be able to use this service won't be a large area of concern for us. 

Our full time staff are Dev Shah, Hazel Bains, Aryan Khimani, and Anya Singh. All of our profiles are linked on our website (mamitahealth.com) and have details about our backgrounds.

Overall, our team is extremely skilled in the field of technology. Two of our full-time staff are established in the field of Artificial Intelligence and have projects in areas ranging from Natural Language Processing and Understanding to Neural Network Training. With an extensive background in tech, we bring knowledge and perspective from fields like Blockchain and Gene Editing. 

As a team, we're deeply committed to the mission of Mamita Health. Having had experiences working in hospitals, we've seen firsthand the intricacies of the healthcare system in Canada. While exploring issues relating to healthcare, as a team, we realized, there was a huge gap with maternal awareness and education not just in developing countries but many places around the world. 

Personally, I've seen the kinds of struggles many women in my life go through when they leave their busy lives to take care of themselves and their children. In western countries, the importance of maternal healthcare is emphasized heavily. Many women, unfortunately, lack the resources to even know how to take care of their children and themselves during pregnancy. I realized that even answers to small questions like "is this normal?" are extremely valuable for keeping peace of mind about the safety of a pregnancy. This was the inspiration behind our solution. To be able to give access to the quality maternal education that many mothers in western countries take for granted is what we want to do with Mamita Health. We believe that by educating women, we can make an impact and save millions of lives. 

Our solution will lessen the burden on the healthcare system in Bolivia if women are cared for in an organized and timely manner. The solution solves the problem of sustainable healthcare in select developing countries where lack of education causes many issues as a result of the culture and accessibility in a country. Contributing to creating a more sustainable and accessible healthcare system in Bolivia will not only improve outcomes for women, but also the rest of the Bolivian population. The team is thoroughly committed to the role that sustainability plays in issues relating to making healthcare accessible, affordable, and of course, sustainable. 

The $10,000 the HP Girls Save the World Prize will grant will be put towards initiating the pilot in Bolivia. Currently, with a lack of partners and sponsors (although we're looking for support from Twilio and MSI Choices), we have no other means of accessing funds as none of our team has a stable source of income to invest in the project. The HP Girls Save the World prize would empower our team to help save the lives of millions of women around the world who could benefit from our solution. 

The Pozen Social Innovation Prize is for solutions addressing unmet needs relating to women and girls. Our solution is targeted at pregnant women in developing countries. We designed our solution for Bolivia which is one of the poorest countries in the western hemisphere. Our pilot is aimed at indigenous women in Bolivia and is planned to run in the regions around Lake Titicaca. Women, especially in developing countries, are often unrepresented at every level. There's an overall lack of solutions addressing the lives and challenges of women and girls. Many countries like Bolivia, lack proper maternal education. Our solution fills this gap by making remote maternal healthcare more accessible, effective, and inexpensive. With the help of the Pozen Innovation Prize we will have enough funds to make this project a reality and make an impact on the lives of women who continue to be left out of key decision-making around solutions in healthcare.