Disease Surveillance Made Easy

Integrating social media chatbots with DHIS2 to fast-track latest case reporting, numbers and investigation in real-time to aid individuals and government decision-makers better respond to disease outbreaks to protect the health of their populations.

Joseph Mulabbi, George Mwangi Maina.

In order to effectively respond to disease outbreaks and contain epidemics, timely detection and containment of cases are crucial, yet routine disease surveillance systems often lack efficient and scalable reporting tools. The ongoing COVID-19 pandemic has laid bare obvious gaps in disease surveillance, particularly in the private sector, which is the often the first point of care for people seeking fever treatment. For example, an estimated 75% of people in Uganda and 90% in Uganda’s capital of Kampala first seek care for fever in private facilities, confirming the need to further invest in surveillance within this sector. This is particularly the case in malaria, Rubella elimination settings, where standard protocols require every case to be reported within 24 hours to the local response teams and national health authorities.

Digital reporting solutions aimed at replacing traditional paper-based reporting mechanisms have been introduced in Uganda just like in many countries across Africa and globally, but these are generally dependent on substantial investments in training of end-users and in equipment procurement. Even when financial and human resources are available, many mobile data collection solutions present challenges with software maintenance, device compatibility, and user management, and are thus often difficult to deploy at scale in a sustainable manner.

Disease surveillance systems in African countries and globally often lack efficient and scalable digital reporting tools to effectively respond to disease outbreaks and contain epidemics. The COVID-19 pandemic has revealed gaps in disease surveillance particularly in the private sector, which is often the first point of care for people seeking fever treatment.

Test. Treat. Track. These three words guide efforts towards disease elimination—as well as combating other deadly diseases. In many countries, that last step, tracking, is still facilitated using traditional paper-based reporting methods or complicated digital reporting tools that makes it impossible to capture and share critical data rapidly, identify where the disease is occurring, and ultimately curb its spread. Even individuals in any given community lack a reliable platform for accurate health information given the many sources of false information platforms.

Chatbots, Faster & Easier 

In collaboration with partners such as the LivingStone International University, we have tested and currently refining our chatbots built on popular social media platforms, such as Facebook and WhatsApp, to support reporting malaria, Rubella, and Covid-19 cases and other disease surveillance data. This tackles the issue of delays from paper–based reporting and decreases the barriers to entry presented by complicated mobile reporting tools. The chatbots allow front-line health workers from pharmacists to doctors to leverage a social media platform they already use on their own mobile devices, in their local language.  

Our Elimination of Disease Outbreaks through Surveillance project works with private sector providers to increase access to quality disease case management. The project facilitates the reporting of disease case data from the private sector into national surveillance systems.

This easy-to-use messaging service also has the potential to reach 2+ billion people and enables government Ministries of Health and their health partners to get information directly into the hands of the people that need it.

From government leaders to health workers and family and friends, this messaging service has capacity to provide the latest news and information on disease outbreaks such as Malaria, Rubella, Coronavirus including details on symptoms and how people can protect themselves and others. It also provides the latest situation reports and numbers in real-time to help government decision-makers protect the health of their populations.

The service can be accessed by a link that opens a conversation in Facebook Messenger  and WhatsApp. Users can simply type “hi”, hello, to activate the conversation, prompting a menu of options that can help answer their questions about Malaria, Rubella, and/or COVID-19.

Ultimately, the social media chatbots connected to DHIS2 complement traditional reporting mechanisms by providing health care providers with easy-to-use and low-cost alternatives. We expect that wide-scale adoption of chatbots will result in more timely and more comprehensive disease surveillance data that are fully integrated in the national HMIS, rendering outbreak detection and response more efficient.

The new service, which is free to use, has been designed to answer questions from the public about infectious disease outbreaks such as Coronavirus, and to give prompt, reliable and official information 24 hours a day. It helps the Ministry of Health to combat rumors and provide the public with reliable and credible information and practical advice to protect them from the virus.

Our tool is free to use for both individuals and healthcare actors. The chatbot offers a wealth of information to users - including practical advice on how to protect yourself from coronavirus, answers to frequently asked questions, directions for verifying the facts and stopping rumors, practical advice for travelers, as well as any other queries that users want to raise. Several doctors and communicators have been made available to respond directly to any queries.

We are pleased to be able to provide the Uganda Ministry of Health with communication tools that will help answer citizens' questions on Malaria, Rubella, Covid-19 and make reliable and credible health advice easily accessible.

On 11 March 2020, the WHO Director-General rang the alarm bell loud and clear by calling COVID-19 a pandemic. Globally and locally, control and prevention measures have been frustrated by myriad challenges. First, accurate information is crucial, but often unknown, or obscured by misinformation. Second, disease fear and confusion contribute to under-reporting of symptoms. Third, preventative strategies such as hand washing or social distancing are costly to disseminate and enforce. Fourth, infection counter measures (e.g., social distancing and quarantine) are psychologically damaging. For example, the SARS outbreak in 2003 resulted in a mental health catastrophe, in which 59% of patients in a hospitalized cohort developed a diagnosable psychiatric disorder, most commonly post-traumatic stress dis-order and depression. After 30 months, less than half of this cohort psychologically recovered. In this light, the WHO has called for large-scale implementation of high-quality, non-pharmaceutical public health measures to help limit new cases, and safely triage those who may be infected. Normally, resources such as clinician attention or emergency department waiting areas are used at a rate the healthcare system can handle. In a pandemic, the cost of these resources being spent inefficiently or contaminated can be catastrophic.

Since the discovery of the Coronavirus (nCOV-19), white it has become a global pandemic. At the same time, it has been a great challenge to hospitals or healthcare staff to manage the flow of the high number of cases. Especially in remote areas, it is becoming more difficult to consult a medical specialist when the immediate hit of the epidemic has occurred. Thus, it becomes obvious that if effectively designed and deployed chatbot can help patients living in remote areas by promoting preventive measures, virus updates, and reducing psychological damage caused by isolation and fear. We bring present the design of a sophisticated artificial intelligence (AI) chatbot for the purpose of diagnostic evaluation and recommending immediate measures when patients are exposed to infectious diseases like nCOV-19. In addition, presenting a virtual assistant can also measure the infection severity and connects with registered doctors when symptoms become serious.

Information dissemination

During a pandemic, people do not know what to do. Doing too little (e.g., not following prophylactic measures) can increase everyone’s risk of infection. Doing too much (e.g., going to the emergency room for mild symptoms) can overburden the healthcare system, wasting precious resources. Thus, reliable information sources are crucial to prevent a “misinfodemic”: the spread of a disease facilitated by viral misinformation. For instance, during the Zika outbreak in 2016, misleading posts spread faster and were more popular than accurate posts on the large social-media site, Facebook. Because chatbots provide a single answer to most questions, they are able to present concise information from credible sources, which may be less overwhelming than social media or web search engines’ long list of results. This matters because false news spreads online both faster and further than accurate news. Chatbots, in contrast to newspapers and online information sources, can often hear and respond in natural language, improving access for people who cannot read or have difficulty using the internet. They can be available any time of the day to answer questions with up-to-date information, and unlike human experts, can concurrently speak with millions of people at the same time in local languages and dialects.

Symptom monitoring

During a pandemic, both individuals and institutions want to know how and where infections are spreading. Individuals want to avoid getting sick, and institutions such as hospitals or local governments need data-informed policies to increase capacity (i.e., ordering more testing kits) and to plan social interventions (e.g., closing businesses). However, efforts to quickly and accurately gather population level infection rates are stymied by individuals’ fear that disclosing symptoms may harm their professional and social lives. Chatbots may be uniquely well suited for symptom screening in a pandemic because people with stigmatized conditions often avoid seeking health care and education. Prior research suggests people are more willing to disclose sensitive personal symptom information to a chatbot than to a human. This means that people may be more forthcoming with chatbots than other humans, providing timelier and more accurate personal triage and population-level infection rate estimates. Healthcare organizations, large corporations like Apple, Amazon, Facebook, Microsoft, and Tencent, governmental agencies like the CDC, and non-governmental organizations like the WHO have launched or helped develop COVID-19 focused chatbots on platforms available to billions of users, likely with the aim of increasing accessibility.

Behavior change support

The WHO Director-General could not say it loudly enough: “all countries can still change the course of this pandemic,” and must do so by mobilizing people in transmission-reducing behaviors such as hand washing and social distancing. To affect behavior, information must be actionable. Chatbots could fill the gap between knowledge and action through repetition, step-by-step instructions, and by suggesting “tips and tricks” for behavior change (e.g., self-enactable behavior change techniques). In a study of low health literacy patients in a hospital setting, 60% requested additional health information from a chatbot at discharge. In a pandemic, chatbots could offload time-consuming but important behavioral support and instruction from human healthcare workers. Home-based chatbots, like the ones on devices from Amazon, Apple, and Google, may support behavior change by linking users to third-party voice apps through “skills” or “actions.” These additional capabilities allow chatbots to provide services and share information beyond their native programming.

Mental health support

Although global and national health bodies highlight the importance of mental health in a pandemic, COVID-19 mental health needs have reportedly been under-addressed. Front-line clinicians are often not trained in emergency psychological support, and mental health practitioners are in short supply. Short-term, chatbots may mitigate the psychological harm of isolation, even though they cannot maintain human-level conversations. Simply disclosing concerns and receiving emotionally supportive responses can have positive value in some contexts. If effectively designed and deployed, chatbots may lessen the long-term harm of pandemic-related isolation, trauma, and depression. Preliminary evidence shows that chatbots may reduce mental health symptoms, but long-term outcomes are unclear and worthy of future investigation.

On balance, our experience shows that these social media chatbots connected to powerful information management systems such as DHIS2 have the potential to change the way that we collect data. They provide health service providers with additional options to rapidly report disease surveillance data to the government, in a user-friendly and low-cost manner.

The chatbots connected to DHIS2 are highly flexible and customizable for each country context and have the potential to be applied to other simple mobile data collection needs. Besides case reporting, this also includes aggregate reporting of activity reports and stock levels: in Laos, our network of private clinics and pharmacies will soon start reporting monthly volumes of malaria RDTs used along with key stock indicators for essential medicines (so far this process was entirely paper-based, with manual data entry in our DHIS2 instance).

Future iterations will also incorporate reminders for providers who see very few cases or have not sub-mitted reports. Likewise, we may prompt them to submit “zero case” reports as these allow for accurate assessments of completeness of reporting. Automated SMS or email notifications based on case reports will be used to instantly inform local health authorities of notifiable disease events in their area.

Ultimately, the social media chatbots connected to DHIS2 complement traditional reporting mechanisms by providing health care providers with easy-to-use and low-cost alternatives. We expect that wide-scale adoption of chatbots will result in more timely and more comprehensive disease surveillance data that are fully integrated in the national HMIS, rendering outbreak detection and response more efficient.

Over the next 12 months we seek to partner with leading NGOs that work with community health workers to build an adapted version of the chatbot, which will be tested and deployed at the field. Our key objective will be to achieve proof of concept among CHWs and drug shop dispensaries for our solution. This entails testing, deploying and validating our solution together with strong partners on the ground. Our focus will be on Sub-saharan Africa.

We shall also seek to test a specialized version of the Chatbot among a small amount CHWs, drugshop dispensaries and health facilities. We start with roughly 100.000 people to be served by the above-mentioned stakeholders. We will have gained first insights and set up an infrastructure, which will allow us to modify and further adapt the Ada to improve decision making and referrals. In three years we will have validated the model and rolled out the specialized version to thousands of CHWs and drugshop dispensaries, serving a large amount of people in East Africa.  

We want to make it easy for anyone, anywhere to access trusted health information, medical expertise and high-quality care, tailored to their individual needs. Our vision is to empower patients and frontline health workers to take better informed health decisions and to enable access to appropriate and timely health data and care in LMIC. Over the next 3-5 years we seek to reach over 20 Million people, directly as the platform users as well as indirectly through frontline health workers. In addition, we aim to impact lives positively, through the use of patient analytics and aggregated, real-time populations insights, which allow for effective outbreak control.

We also look forward to focus on future iterations to incorporate reminders for healthcare providers who see very few cases or have not sub-mitted reports. Likewise, we may prompt them to submit “zero case” reports as these allow for accurate assessments of completeness of reporting. Automated SMS or email notifications based on case reports will be used to instantly inform local health authorities of notifiable disease events in their area.

We expect that wide-scale adoption of chatbots will result in more timely and more comprehensive disease surveillance data that are fully integrated in the national HMIS, rendering outbreak detection and response more efficient.

Our impact measurements are based on our ability to meet and effectively to the user needs and feedback we got during the first user-testing of the product as seen below.

• Users reported a strong preference for the new reporting mechanism compared to “traditional” mobile reporting tools such as DHIS2-connected data collection apps, especially as the chatbot works through a familiar messaging channel on their own devices.

• The automated prompts in local language make it easy to report data.

• This is particularly relevant in settings where reporting frequency is low, as users don’t need to re-member how to navigate a complex digital form

• Training can be done over the phone (e.g. using WhatsApp) or in person by field staff or during meet-ings with providers.

• Digital reporting mechanisms should include a range of options that are context-appropriate and that allow for providers to use a platform that most easily facilitates timely and complete reporting. While the Facebook and WhatsApp bots are accepted, it is important to note that there is no “one size fits all” solution and that options should be tailored to accommodate other preferences, which may include mobile data collection apps, SMS, or even sending pictures of manually completed paper forms – something that is often forgotten when designing digital solutions.

• The bots do not yet remove the need for in-person case notification to the local health authorities who are responsible for case investigations, but they enable easy and fast reporting. Data are thus more rapidly available and open up the potential to improving the detection and containment of epidemics by central and local surveillance and response teams.

• Large networks of providers still require active user account management by central teams and DHIS2 system administrators, who need to ensure that new users get swiftly enrolled and that the social media user IDs are mapped against the DHIS2 organisation units.

Once we are able to fully meet the above bullet points/concerns as raised by our users, then we shall effectively measure our success.

Finally, impact is and shall also be measured based on number of subscribers, engagements, awards, grants, partners both from the private and public sectors, feedback from users etc.

Successful adoption depends on user behavior, meaning how the technology is incorporated into patient-care workflow at a health facility. Limited tech literacy by end-users, limited access to hardware, power outages, and unreliable connectivity present infrastructural barriers. Reliance on paper, and potentially unsupportive leadership present system-based hurdles. We are also trying to develop a platform that can be used universally while adapted to unique contexts and jurisdictions.

Access to partners: Our solution to build and test a specialized version of the chatbot for CHWs and drug dispensaries will only succeed by forming relevant partnerships with NGOs and foundations, in order to start implementation. The Trinity Challenge network would be of great support to connect us with the its community of investors and well as connect with the Solve Community, so we can explore further collaboration opportunities.  

Financial support as well as technical support are also other crucial areas that we require a serious break through in order for us to move the project forward. For this, we shall continue lobbying for financial support from the government and key health partner NGOs we know. We shall also continue engaging in grants competitions such as The Trinity Challenge to compete so that we may be able to win grants and access a network of both investors and technical support networks that might help to push us forward.

Living Stoner International University

Ministry of Health

Waiba Health Services

Living Water Ministries Uganda

We believe in the power of our technology to create long-lasting impact. Establishing partnerships and developing a network of strong supporters is crucial to achieving our mission. Becoming The Trinity Challenge winner would tremendously advance our work, allowing us to connect with relevant organizations in global health. In addition, mentoring and guidance by the respective experts within the The Trinity Challenge network of founder members would be also of great support. Through the challenge we could increase our visibility and raise more awareness for our work within a broader audience. We are also applying to this challenge for the chance to win equity-free funding which, if gotten, will accelerate our expansion timeline. We are also applying for the opportunity to be part of The Trinity Challenge’s brilliant community committed to social impact. We believe that The Trinity Challenge can give us tactics on how best to hold on to our social impact goals while pursuing our business goals. Being a winner will also give us access to some of the most intelligent minds in the world who can provide feedback on our technology solutions and connect us to partnerships which can help scale our business. 

We also welcome the opportunity to exchange ideas and lessons learned with colleagues and peers who have been experimenting with new technology in low resource environments. We value working in partnership and the mutual growing and learning that comes with collaboration. 

  We shall seek partnership with Google, Bill and Melinda Gates Foundation, Facebook etc and this partnership shall solely be for the following goals.

• Peer-to-Peer Networking
• Organizational Mentorship
• Impact Measurement Validation and Support
• Media Visibility and Exposure
• Grant Funding
• Develop prototypes of the chatbot tool adapted for use among different types of semi-skilled health workers
• Introduce an effective referral system across stakeholders within primary healthcare
• Form partnerships with relevant stakeholders, such as local governments and the private sector, to implement chatbot tools on the ground in sub-Saharan Africa and South Asia