LifeBot

For many of the pregnant women and mothers of newborns, the search for healthcare information is challenging in Bangladesh. Many of these women do not know where to go and who to ask for help. Often, this information is not easily accessible. Engagement of a large human resource is required to provide with this information under the current infrastructure of healthcare information services, which may be expensive. Furthermore, people desire for rapid access to information with the engaging and personal interactions with the service providers.

Connecting all these dots, provision of authentic information and continuous healthcare support is possible with minimized cost by an Artificial Intelligence based ChatBot for healthcare services. We would call it "LifeBot".

A ChatBot personality can provide with the immediate personalized services to the targeted pregnant women and new mothers and create an interaction like a personal assistant by building a relationship with them.

Healthcare services in Bangladesh have always been challenging in terms of providing quality services for pregnant women and new mothers for being understaffed and underfunded. According to World Health Organization, there are 3.05 physicians per 10,000 populations and 1.07 nurses for 10,000 populations in Bangladesh.  According to Bangladesh Maternal Mortality and Health Care Survey (BMMS) 2016, the Maternal Mortality Rate (MMR) estimate is 196 maternal deaths per 100,000 live births. Although government is trying to invest and provide support to these women, the need for further support is still existent. Pregnancy and early motherhood are the most vulnerable stages for women when they have plethora of information about their physical and mental well-being most of which are not reliable.

A ChatBot using Bengali language can be drawn as a smart solution to address this gap. Globally, many advantages of ChatBots within the context of health care have been theorized, such as improved patient education and treatment compliance. However, little is known in Bangladesh context how this Artificial Intelligence can help people when they are in need of health care advice. The administrative benefits of a ChatBot include but not limited to, scheduling doctor appointments, locating health clinics providing medication information.

This project will focus on the pregnant women and the mothers of newborn to provide with quality health care information in their native language when required. The pregnant women and new mothers will be able to call a local operator to ask their health related queries to an automated ChatBot. In Bangladesh, most women are still in vulnerable situation when they are unable to seek appropriate health care when required. LifeBot can come in their rescue. LifeBot will have a knowledge base of the Frequently Asked Questions by the pregnant women and new mothers from the previous call centre database. Moreover, a human operated call centre will also be initiated under this project to enrich the knowledge base for LifeBot. ChatBots can be benefitted over call centres by reducing the burden of human resource and for its ubiquitous availability.

icddr, b is an international research institute based in Bangladesh proposes a data bank with reliable and valid maternal and child health information in collaboration with a2i – Access to Information (Prime Minister’s office of Bangladesh) to provide with the capacity of developing high-level programming on AI. a2i will help icddr,b to facilitate the feasibility study in their established Union Digital Centres (UDC) in rural areas upon receiving the fund.

This project will develop the ChatBot named LifeBot to help pregnant women and mothers with newborn. Alongside, it will also conduct a feasibility study to understand the effectiveness and safety measures of this programme through a user test. A pre-post intervention study will be conducted on a small sample in order to understand the efficiency and usefulness of the ChatBot.

We expect to obtain a ChatBot programme for providing health care advice to pregnant women and mothers with newborn and the recommendations and user guideline on this ChatBot system for ensuring the safety of the users. Specifically, this will focus on the following objectives:

·       To develop a ChatBot named LifeBot to provide with the essential health care knowledge to Bangladeshi pregnant women and mothers with newborn for their physical and mental well-being

·       Conduct a pilot study (user-test) to assess the effectiveness and safety of the ChatBot.

The process will continue in four different stages:

1.            Develop the knowledge base using call centre communication data

2.            Developing the AI agent to make it able to understand different voice inputs

3.            Feasibility test in the Union Digital Centres (UDC) managed by a2i

4.            Scaling up the ChatBot service to Upazila Health Complex and District Hospitals

Initially, a prototype will be attempted to be developed by collating the call center’s database to generate the knowledge base. However, we anticipate some challenges to develop a ChatBot in Bengali language because the AI need to adapt and learn to understand different intonations and this will be time consuming. Therefore, once the basic knowledge base is developed, this project will train the AI agent to interpret the Bangla queries with improved quality. After that, this programme will be tested on the users of Union Digital Centres (UDC) of Bangladesh managed by a2i by conducting a feasibility test. Later, this service once successfully launched will be scaled up with the government support. 

Quality of care is important to be considered in health service delivery to pregnant women and mothers of newborns. In developing countries like Bangladesh, the number of doctors per patients is not enough which leads to long waiting hours, lack of proper consultation during visits and lack of proper communication such as anger during visits. However, it is very important for a mother to have access to pregnancy, delivery, and newborn health related any information as well as support for their mental well-being.

Including ChatBot in the health service delivery system of developing countries like Bangladesh will address most of the issues the countries have been facing. ChatBot can provide faster and immediate responses to the patients. The patients can raise their concerns anytime just by calling the ChatBot numbers. Based on the problems shared, the ChatBot can immediately refer the patients to proper referral facilities.

Our proposed LifeBot is innovative in Bangladesh to improve patients' outcomes and would minimize the health related costs of the pregnant women and new mothers.

ChatBots that uses powerful Artificial Intelligence-driven algorithms could help patients in their native language for miscellaneous health services. This can start from choosing a doctor and finding the nearest health facility to emotional supports. Moreover, these services would be available all the time. ChatBots can even help patients with prescription, appointment, therapy reminders, or post treatment follow-up.

We expect our proposed LifeBot would be able to support the pregnant women and new mothers on their physical as well as mental health issues and queries. icddr,b will provide them with the reliable information. Furthermore, a2i would attempt to roll this project with Bangladesh government's support after developing and testing the prototype in a smaller scale.

This project is still undergoing its concept development and this is not currently under use. However, if successfully launched LifeBot is expected to serve all pregnant women and mothers of newborns in Bangladesh in the upcoming years.

In next 1 year, we expect to develop the prototype of LifeBot for healthcare information services in Bangladesh. Considering various Linguistic pronunciation in our country, an android app can read the words or sentences or commands of a user using various “deep learning algorithm” which convert the speech to text. Language Identification (LID) systems are used to classify the spoken language from a given audio sample and are typically the first step for many spoken language processing tasks, such as Automatic Speech Recognition(ASR) systems. By recognizing the pattern using sound wave that algorithm store the approximate solution to text. This text can be analyzed as command and according to that command the system will find the referral address or link or number from the list of logic for different command inside the system. After getting the expected link the system will redirect to that referral number. Before that the system will read out the solution to the user by system generated voice. A Text to speech synthesizer is a system that converts text into spoken word, by analyzing and processing the text using Natural Language Processing (NLP) and then using Digital Signal Processing (DSP) to convert this processed text into synthesized speech representation of the text. This process will undergo within the next year. 

Within the next five years, this project will be scaled by Bangladesh government supports. Moreover, this platform will be shared with other contributors who can provide programming or health knowledge support by investing on this project.

Every human is unique individual and their command over language such as using words for love, tension, anger, frustration, depression, happiness, and other emotions is unique and can very indefinite times. Therefore, designing a chatbot like a human who has the ability to deal like a human requires continuous effort and time. We assume this will be the biggest challenge and barrier to develop this ChatBot in native language. Rigorous programming skill would be required to generate this algorithm of language understanding of this automated system. 

With the support of a2i's Artificial Intelligence team we will attempt to improve the Bengali language understanding of LifeBot. The knowledge base would be continuously updated and the AI will be expected to have an automated learning system.

icddr,b and a2i will jointly work on this project. 6 full time staff members of icddr,b will work to develop the knowledge base and a team of 10 full time Artificial Intelligence programmers will work from a2i. This platform will be open for collaboration after developing the prototype. 

This project is being proposed by icddr,b, located in Dhaka, the capital city of Bangladesh, is one of the world’s leading health research institutions. Established in 1960, icddr,b has developed and delivered low-cost, scalable solutions to major public health problems— such as oral rehydration solution (ORS), zinc for diarrhea treatment, tetanus vaccines for newborns and newborn care, – which saved and continue to save millions of lives globally.

Integral to its research, icddr,b provides high-quality care, free of charge, to over 200,000 Bangladeshis annually. icddr,b has worked with over 100 organisations including Government of Bangladesh (GoB), other national governments, international institutions and national and international civil society partners and NGOs. And one of icddr,b's greatest contributions lies in its diligent dissemination and scale-up of findings and discoveries through this large network. A knowledge base on health sector developed by icddr,b is expected to be unparalleled by all means.

Our collaborating partner Access to information in Bangladesh or a2i is a UNDP and USAID-supported Access to Information Programme operated from the Prime Minister's Office of Bangladesh is the pioneer of building strategies of Artificial Intelligence for Bangladesh. a2i can support in scaling this project for longer term as it directly works with governments. The sustainability of a project cannot be ensured if it is not rolled out by any government initiative. 

a2i had plans to introduce Artificial Intelligence Task Force in Bangladesh which will be acting as government authority to govern AI implementation in Bangladesh. This Task Force will motivate to administer regulation, implementation and maintenance of Artificial Intelligence. 

Collaboration between these two organizations is expected to have the most impactful results in terms of developing a chatbot for healthcare services for pregnant women and new mothers in this country.

icddr,b and a2i jointly propose to develop this chatbot service called LifeBot. After creating the knowledge base platform it will be made open to other contributors. Organizations that work on Artificial Intelligence programming or health sectors will be able to invest on this platform to improve the system. The database will earn revenue by sharing the information with other potential contributors.

If this funding is obtained, a2i and icddr,b will form a Memorandum of Understanding to work further on it. Considering the potential of this funding the pilot project to build the knowledge base and a conversational AI agent is being proposed. After the successful inauguration of this chatbot project, this will be further scaled by government initiatives with the help of a2i depending on the availability of further funds.

Tiger Challenge is unique in terms of collecting the technological solutions to ameliorate the health sector services. The feasibility of a Chatbot service in health is still largely unknown in Bangladesh. This platform seems to be the best to explore the potential of such project and obtain the grant to build the prototype LifeBot.

We would like to partner with the organizations who can provide more supports to enrich the health knowledge data base and improve the Artificial Intelligence programming system during further scaling of the service.