Respir

220 million. That is Nigeria’s current population and out of that number as of 2023, only 5.7 million individuals have been tested for Covid-19. The advent of Covid-19 engendered a global shock on the world. As a result, many developing countries like Nigeria have faced difficulties related to Covid-19. From economic downturns to social dilemmas the problems are truly profound. However, one of the most prominent issues which plagued Nigeria was its inadequate testing capacity for Covid-19. This was further exacerbated by Nigeria’s already under-resourced healthcare sector where the average ratio of doctors to patients is 1:10,000. Even though PCR and antigen tests have been made available to the Nigerian public, these $137 tests remain inaccessible and expensive for the masses with more than 40% of Nigerians living below the poverty line (earning less than $1.90 a day) according to the National Bureau of statistics. Furthermore, these traditional testing methods are also time-consuming, taking up to 24 hours for results to be delivered to the patient. Moreso, the uncomfortable and invasive nature of the latter testing methods has discouraged many individuals from the experience, resulting in many Covid-19 cases going undetected and mass hysteria being spread.

Even though we are slowly morphing into a ‘new normal’, it is essential to remember that SARS-CoV-2 still poses a severe threat, towards the elderly population who, according to Kaiser Family Foundation, “account for 75% of all COVID deaths globally to date”. The situation in Nigeria is dire and as a result, urgent action is necessary to contain the spread of the virus and protect the most vulnerable populations. A, fast, low cost, and accessible testing alternative is therefore needed to alleviate the burden of the pandemic on healthcare workers and to ameliorate the testing capacity of developing countries like Nigeria. This way, more active cases would be detected on time and contained.

Introducing RESPIR- the ultimate game-changer in the fight against Covid-19. By harnessing the power of artificial intelligence and machine learning, Respir offers a fast, affordable, and accessible solution for screening the presence of Covid-19 using cough sounds. Within just a minute, Respir can screen 10-15 different cough samples. Our product availability spans smartphones, smartwatches and the web, to maximize its accessibility as this is one of our primary missions. Gone are the days of waiting in long lines for a Covid-19 test or worrying about the cost and accessibility of healthcare. With Respir, individuals can quickly and easily assess their health status from the comfort of their homes.

We built this AI model using Python and visualized our code using Jupyter Notebook. We also ensured to integrate our trained model into a web application which we built using HTML, CSS and JavaScript. As this is an artificial intelligence model, it relies heavily on not just the quantity but also the quality of the dataset used to train the model. This is why our model was trained using one of the largest expert-labelled cough datasets in existence, ‘COUGHVID’, which comprises 25000 crowdsourced cough recordings representing different genders and geographic locations. Furthermore, over 2800 of these recordings were labelled by four experienced physicians to detect the abnormalities in the cough recordings.

When users open the Respir app, they are greeted by a disclaimer which informs them about the nature of our app before usage. Users are then able to upload their cough sound into our application alongside other supporting diagnostic questions such as ‘muscle pain’, ‘fever’, and ‘gender’. The Mel-Frequency Cepstral Coefficients of the cough sounds are then extracted. These extracted features are then fed into our 5-layer artificial neural network created by Keras which Is trained to recognize patterns in the MFCCs that are indicative of Covid-19 positive, symptomatic, or negative cough sounds. This classification is then concatenated with demographic features previously inputted. In a matter of seconds, the user is given a prediction of their covid-19 status as either ‘healthy’, ‘symptomatic’ or ‘Covid-19’. Users are also given more information about Covid-19 and are provided with the option to connect to medical professionals in real-time, in the event that they test positive for Covid-19.

Respir aims to serve a target population which is split into three distinct categories: Medical professionals, senior citizens, and adults. With the abrupt nature of the pandemic, the mental health of many medical professionals in local hospitals and designated test centers plummeted as they felt the overwhelming outflow of patients into testing centers during the pandemic. With the already existent issue of brain drain in Nigeria where about 50 doctors leave Nigeria every day, our healthcare system is in urgent need of support, and this is what Respir aims to offer.

Undoubtedly, out of all the other demographics, senior citizens are the most vulnerable to Covid-19. According to frontiers, about 80% of deaths due to Covid-19 occur in those over the age of 65. Many senior citizens are homebound, meaning that they are usually unable to commute to testing centers. As a result, they often remain oblivious to their Covid-19 status until it worsens. In response to this, our solution would provide a remote testing alternative for senior citizens removing their need to travel to testing sites. As opposed to traditional testing methods which often take about 24 hours for results to materialize, Respir provides test results in a matter of seconds allowing seniors seek medical attention promptly through our integrated telemedicine feature if they test positive for Covid-19.

Moreover, many seniors in Nigeria suffer elder abuse and neglect as there are little to no functional senior homes available here. In addition to this, the family members of these senior citizens are often plagued with the worries of their 9 to 5 jobs, neglecting the health and wellbeing of these senior citizens, especially the poor and disadvantaged ones. One of Respir’s main aims is maximum accessibility whereby our application would be available to seniors living in remote or rural areas. This would be done by providing a low-cost physical device which would serve as an offline alternative to our AI application. This way hospitals and individuals residing in areas without strong internet connection would still have access to our application.

Our third target population consists of adults. After surveying 131 people from 6 different age brackets and 20 countries, we found that 62.4% of respondents found their testing experience uncomfortable. From the awkward nasal swabs to the painful blood tests, it is no doubt that traditional testing methods are invasive and uncomfortable for many. However, Respir offers a non-invasive and pain-free testing method for these adults. Our survey also revealed that 89.3% of respondents found traditional testing methods unaffordable for the masses in their respective countries. With the average price of PCR tests being about $137 compounded with the fact that over 40% of Nigerians are living below the poverty line, it is clear to see that these testing methods are costly for an average Nigerian adult especially for uninsured individuals. Hence, Respir is offering a more cost-effective alternative for adults which we believe would play a great role in tracing new cases and controlling the spread of Covid-19.

Having lived in Nigeria for many years, our team has had a first-hand experience of the inadequacy of Nigeria’s healthcare system. From seeing many people in our community suffer from the medical negligence of many underfunded public hospitals to experiencing the effects of Nigeria’s minute annual budget towards healthcare which makes up only about 5.35% of our National budget, it is clear to see that our healthcare system is actively regressing.

This issue is one which has persisted in Nigeria for years and due to this the Team lead, Emmanuela and her teammates Ifunanya and Chioma have developed a passion for the amelioration of Nigeria’s healthcare sector. To ensure that our app meets the needs of our target audience, we carried out extensive field research in Lagos state. We interviewed medical professionals, senior citizens, and adults to understand the severity of Nigeria’s inadequate Covid-19 testing capacity. We found that many medical professionals were initially overwhelmed by the number of patients flooding into the hospital during the pandemic. We also discovered that many senior citizens and adults had superficial knowledge about Covid-19, which deterred them from getting tested. This valuable feedback helped us make necessary improvements such as the addition of the information desk feature providing users with valuable information on the Covid-19 pandemic.

In addition, we surveyed 131 people from 20 countries and 6 different age brackets to understand the needs of a broader population outside of Nigeria. This survey allowed us to find our potential users’ pain points and wants, and we received helpful feedback that prompted us to add more features to our app, such as the telemedicine and language translator features.

Our team lead, Emmanuela, also travelled to one of the rural areas in Nigeria, particularly her village in Cross-Rivers state, and found that the locals were greatly underserved. This prompted us to model a low-cost physical device that does not need an internet connection to operate. This way, we can provide adequate healthcare services to hospitals and locals residing in remote areas. Furthermore, Emmanuela is currently carrying out remote data science research with a professor at the University of Florida to find the most efficient methods to apply data science to the field of health.

As we continue to test our product, we are keeping our target users in the loop to ensure that our app is fully serving their needs. We are committed to designing and delivering a solution that takes into consideration our community's input, ideas, and agendas. Our team is confident that we have the expertise and proximity to the communities we are serving to make a meaningful impact on Nigeria’s healthcare sector with our app, Respir.

Our solution ‘Respir’ is currently a prototype as we have spent months completely building and training our AI model which is now functional. We have also deployed our model into a user-friendly web application which is available online. Having built a working prototype of our app, we have tested the app in 2 local hospitals and are still testing the app amongst a broader population from different demographics. So far, we have received feedback from many doctors and nurses on the operation of our application to which we made necessary improvements. Many doctors suggested that other symptoms should be included in the application other than the cough sounds as this would improve the accuracy of our model. We added more symptoms to our application such as ‘fever’, ‘headaches’ and ‘past respiratory illnesses’ in response to this. Many professionals also raised concerns about the privacy of their data to which we added a disclaimer feature to our application which provided clear and transparent consent mechanisms, explaining to users how the app collects and uses data.

During this prototyping stage, which we are in, we are still making changes to the application ensuring to focus on the user experience, design, and functionality of the app. Moreover, we have modelled a low-cost physical device using Computer-aided design and are working on deploying our AI model into the physical device. Many of our test users also highlighted how they would love to have text-to-speech functionality for visually impaired individuals which we are currently developing now.

Respir aims to cater for the needs of its users which is why we are in constant communication with our target users. We always ensure to inform them about any current developments to our application before making it to ensure that our app is tailored towards their specific needs.

As we are still in the prototyping phase, Respir is not yet serving anyone. Given the importance of accurate medical screening and diagnosis especially towards a novel virus like SARS-CoV-2, we are working hard to develop and refine our app so that it is accurate, reliable, and effective. We are working closely with medical professionals and data scientists to incorporate their feedback and improve our app’s reliability. We understand that Covid-19 is a serious public health issue that affects millions of people worldwide and we are committed to doing our part to help combat the pandemic, but we also recognize that it is essential to ensure that the app is thoroughly tested and validated before it is made widely available.

We are confident that once Respir is fully tested, developed, and launched, it will have the potential to serve millions of people worldwide who are affected by the inadequate Covid-19 testing capacities of their respective countries.

Upon reflecting on our application to this challenge we soon realised that Solve is not just any other competition, it is a hub of innovation and a centre which stimulates breakthroughs. We believe that the Solve community is the perfect place to develop our solution as the unwavering support offered by this community is like no other. From the coaching and advice offered by Solve experts to the interconnected and interdependent nature of networks at MIT, applying to solve would be a key to enhancing our solution.

We believe that applying to this challenge would expose us to a plethora of experienced techpreneurs who can serve as mentors to our team and help scale our solution to the best version it could be. Emmanuela, our Team Lead has a burning passion for technology and its intersection with health and so do her teammates. We believe that the Solve community would connect us with like-minded individuals with similar interests who would be able to make necessary suggestions and contributions to the development of our solution.

More specifically, as we are a startup, we believe that the grant provided by Solve would be very valuable towards our app’s development. As our app is based on an artificial intelligence model, the quantity and quality of data used to train the model are key. We plan to invest a percentage of the grant into conducting further research and acquiring a bulkier dataset to train our model to greater accuracy. We also believe that Solve’s diverse network will connect us to data scientists, Artificial intelligence engineers and medical professionals who we can team up with to further develop our solution.

 As medical data is classified as ‘sensitive personal data’, it is essential that we prioritise the privacy and security of our users’ data. Covid-19 is a global issue as it affected virtually every country around the world. These different countries have different data protection acts which govern how personal data should be collected, used, stored, and shared. Applying to solve would give us vital support and provide necessary resources on how to handle the legal aspect of our solution, connecting us with a pool of experienced individuals in the legal field. This way we will be able to develop, test and deploy our solution in the most effective manner.

Furthermore, we recognize that cultural and market barriers also exist. The healthcare landscape in Nigeria is complex, with different cultures and beliefs impacting how individuals perceive and seek healthcare. We believe that MIT Solve can help us understand these cultural and market barriers and provide us with the necessary guidance to navigate them effectively.

We firmly believe that Respir presents a ground-breaking approach to Covid-19 screening, one that has the potential to transform the way we detect and mitigate the spread of this deadly virus.

As opposed to the traditional testing methods which involve invasive nasal swabs and long, strenuous hours of testing, one of Respir’s value propositions is its speed and real-time analysis. Respir’s complex network of deep learning algorithms and artificial neural networks allows it to process inputted cough sounds and produce outputs in a matter of seconds, making it the fastest Covid-19 screening method compared to traditional testing methods. Furthermore, Respir’s model is continuously learning as more data is fed into the system as users make use of the app. This way we are constantly improving the accuracy and reliability of our model.

Furthermore, Respir offers multilingual support and text-to-speech translations which shows that our solution caters for a diverse range of people from different backgrounds. Our privacy-focused design also sets us apart from the competition as we ensured to maintain the privacy and security of our potential users’ cough data. Our key differentiator is Respir’s affordability being 96.35% cheaper than traditional testing methods. We believe that our price sets us apart because it opens testing for people in multiple socioeconomic backgrounds as opposed to the expensive tests which were only accessible to middle to upper-class members of society.

We are confident that Respir has the potential to spur broader positive impacts by inspiring collaborations between researchers in healthcare as well as individuals in the field of Artificial intelligence. This integration of technology into the health sector could potentially lead to more innovative early-detection solutions for other respiratory illnesses. Moreso, Respir could equally encourage greater international cooperation and information sharing in the fight against Covid-19 particularly in regions with limited healthcare resources.

In terms of changing the market, Respir has the potential to disrupt traditional diagnostic methods and ultimately transform the way that we currently approach public health. Our solution has the potential to become a standard tool for healthcare providers and public health agencies around the world and due to its technical nature, Respir is very scalable as opposed to traditional testing methods. We also believe that Respir could support many medical professionals making their work more conducive to do. In terms of cost, Respir’s inexpensive nature could lead to cost savings for both healthcare providers and patients, ultimately encouraging more individuals to invest in their health and well-being.

Impact Metrics

One-year target

• Partner with 10 hospitals in Nigeria to deploy, test and improve the accuracy of our solution, reaching more than 20,000 patients.
• Conduct clinical trials in regions with high Covid-19 infection rates and work with local healthcare providers to ensure accurate data collection to improve the accuracy of our model.
• Build and test a low-cost physical device alternative and deploy this device in 3 local hospitals in rural areas without internet connection to test its reliability and speed of screening.
• Expand Respir’s functionality such as integrating remote monitoring of patients by doctors.
• Establish a training program for healthcare providers on the use of the Respir app with the aim of training at least 1,000 medical professionals.
• Collaborate with research institutions and data scientists to further develop and improve Respir’s accuracy and effectiveness.

Plan of Action for One-Year Target

One of our team members, Chioma, has numerous connections with different hospitals across Nigeria due to many internships she has participated in. With these connections as well as the connections provided by solve, we believe that partnering with hospitals in Nigeria would be simplified. We plan to research on these hospitals and identify hospitals in Nigeria that have a high volume of patients and are in areas with high Covid-19 infection rates. We then plan to contact hospital administrators and schedule a demo of the Respir app for the hospital staff including doctors, nurses and lab technicians showcasing the app’s features and functionality. This would then prompt us to provide training for these hospital staff including hands-on experience with the app. We will also ensure to monitor the usage and efficacy of our app in partner hospitals. We are choosing the deploy the app first in partner hospitals rather than in the general public as these hospitals provide a more controlled and observable environment for the development and improvement of our application. With regard to the development of the low-cost device. We plan on building a first prototype which would be tested in 3 local hospitals in rural areas. We will only build 3 prototypes to identify any flaws in these prototypes and hence make necessary improvements. To further develop Respir’s effectiveness, we will use the Solve network to access any data scientists or machine learning experts at research institutions who can give their necessary input on how to further improve our product.

Five-year target

• Expand the reach of the Respir app to more than 100 countries and 50 million users worldwide including developing nations where access to Covid-19 testing is limited.
• Achieve a 90%+ accuracy rate in diagnosing Covid-19 cases using the Respir app
• Secure partnerships with at least 50 hospitals and healthcare organizations worldwide to integrate the Respir app into their healthcare systems.
• Develop and launch a version of the Respir app that can diagnose a wider range of respiratory illnesses beyond covid-19.
• Engaging with regulatory bodies and policymakers to ensure that RESPIR is compliant with relevant laws and regulations
• Increase the number of healthcare providers trained on the use of the Respir app by at least 20,000.

Plan of action

We plan on developing relationships with global healthcare organizations and academic medical centres to expand the reach of the Respir app. To achieve 90%+ accuracy we will ensure that our application is continuously learning. Therefore, as more users input data into the model, the more accurate it becomes over time. Furthermore, we will implement the Training of trainers (ToT) model to ensure continuous and quality training of medical professionals in the use of our app, Respir. 

To measure our progress towards our impact goals for Respir we are using the following indicators:

• The number of users – we will track the number of healthcare providers, public health agencies and individuals using Respir. This way we can monitor and evaluate the reach of our solution.
• Accuracy and specificity- We will continuously evaluate RESPIR's accuracy and specificity in detecting Covid-19 by carrying out multiple clinical trials.
• Time to diagnosis- We will measure the time it takes to diagnose Covid-19 using Respir and aim to continually reduce this time to ensure early detection and treatment. This way we will ensure that one of our value propositions, ‘speed’, would be maintained.
• Geographic coverage- We will also track the geographic coverage of Respir including the number of countries and regions where it is deployed, making sure that our solution is accessible to people of different backgrounds as well as people in underserved communities.
• Reduction in disease transmission- We will aim to measure the reduction in Covid-19 transmission rates resulting from the use of RESPIR in clinical and public health settings.
• Cost savings-We will track the cost savings achieved through the use of RESPIR compared to traditional diagnostic methods, to demonstrate the value of our solution in improving healthcare affordability and accessibility.

We are also taking into consideration the target indicators associated with UN Sustainable Development Goal 3: Good health and well-being.

• Target 3.3: By 2030, end the epidemics of AIDS, tuberculosis, malaria and neglected tropical diseases and combat hepatitis, water-borne diseases and other communicable diseases.
• Target 3.8: Achieve universal health coverage, including financial risk protection, access to quality essential health-care services and access to safe, effective, quality and affordable essential medicines and vaccines for all.
• Target 3. c.1: Substantially increase health financing and the recruitment, development, training and retention of the health workforce in developing countries, especially in least developed countries and small island developing States.
• Target 3.d: Strengthen the capacity of all countries, in particular developing countries, for early warning, risk reduction and management of national and global health risks.

• Indicator 3.3.2: Tuberculosis incidence per 100,000 population
• Indicator 3.8.1: Coverage of essential health services
• Indicator 3. c.1: Health worker density and distribution
• Indicator 3.d.1: International Health Regulations (IHR) capacity and health emergency preparedness

SDG 17: Partnerships for the Goals

Target 17.16: Enhance the global partnership for sustainable development, complemented by multi-stakeholder partnerships that mobilize and share knowledge, expertise, technology and financial resources, to support the achievement of sustainable development goals in all countries, in particular developing countries.

• Indicator 17.16.1: Number of countries reporting progress in multi-stakeholder development effectiveness monitoring frameworks that support the achievement of sustainable development goals.

By tracking these SDG targets and indicators, we can demonstrate the impact of RESPIR in contributing to the achievement of global health and development goals.

Outline of our Logic model

Problem statement- The Covid-19 pandemic has revealed the gaps in many health sectors in developing countries around the world that battle with inadequate Covid-19 testing capacities for their citizens.

Desired state- A world where there is a fast, inexpensive, accessible, and reliable Covid-19 screening tool which makes Covid-19 testing more convenient for patients and healthcare professionals.

Assumptions:

- Healthcare providers and individuals will be willing to adopt and use the Respir app as a primary diagnostic tool for COVID-19.

- Respir will be able to diagnose COVID-19 cases with high accuracy and reliability, even in populations with varying cough sound patterns.

- The Respir app will be accessible and affordable for healthcare providers and individuals in both developed and developing countries.

Inputs:

• Skilled development team and AI experts
• Access to relevant data sets of cough sounds
• Adequate funding and resources to develop and deploy Respir.

Activities:

• Develop and train artificial neural networks to detect patterns in cough sounds that may indicate the presence of COVID-19
• Continuously test and refine Respir using real-world data.
• Collaborate with healthcare providers to conduct clinical trials and research studies to validate the effectiveness of the Respir app.
• Provide comprehensive training and support to healthcare providers and individuals on how to use the Respir app effectively.
• Continuously gather feedback from healthcare providers and individuals to identify areas for improvement and make necessary updates to the Respir app.

Outputs:

• A fully functional and effective version of the Respir app and low-cost physical device that can diagnose COVID-19 cases with high accuracy and reliability.
• A network of partnerships with healthcare organizations and NGOs to distribute the Respir app to populations in need.
• Comprehensive training and support materials for healthcare providers and individuals on how to use the Respir app effectively.

Short-term Outcomes:

• Increased access to COVID-19 screening, particularly in low-resource settings where traditional diagnostic methods may not be readily available.
• Early detection of COVID-19 cases, leading to better outcomes for patients and a decrease in transmission rates
• Improved efficiency and effectiveness of healthcare systems by providing a fast and reliable screening tool.

Long-term outcomes:

• Reduction in the spread of COVID-19 and other respiratory diseases, leading to improved global health outcomes.
• Increased access to healthcare, particularly for marginalized communities and individuals in low-resource settings
• Improved capacity of healthcare systems to respond to outbreaks and pandemics.
• Advancement of AI-based diagnostic tools and machine learning algorithms, leading to improved healthcare outcomes in other areas

Our survey of 130 individuals from 20 countries revealed that 83.2% of respondents preferred our app to the traditional testing methods. Also, according to a research paper on ‘Covid-19 self-testing’ (https://doi-org.ezproxyberklee.flo.org/10.1371/journal.pone.0282570), “While just a few informants were familiar with SARS-CoV-2 self-testing, they generally supported using self-testing. These are key success indicators that target audience would be willing to accept our product which would help us achieve our desired short and long-term outcomes.

Respir is an application which is powered by artificial intelligence and machine learning. Our application harnesses deep learning techniques to screen for Covid-19 using cough sounds. The main programming language used to build our model was Python programming language. We also made use of Jupyter Notebook in building this solution as it provided an environment for us to not only write and execute our code but also to visualize our data all in one place. To make our model useful we deployed it into a web app which was developed using HTML, CSS and JavaScript.

For our data, We made use of the Coughvid data set which we sourced from nature.com. This dataset provides more than 25,000 crowdsourced cough recordings which represent a wide range of participant ages, genders, geographic locations as well as Covid-19 statuses. 2,800 recordings from this dataset were also labelled by four experienced physicians to diagnose the abnormalities present in the cough sounds. We picked this dataset because it is one of the largest expert-labelled cough datasets in existence and hence can yield more accurate results.

After obtaining our dataset, we proceeded to build our Artificial intelligence model in 5 main steps: Loading the data, pre-processing, training the model, evaluating the model, and testing the model.

The first step was data loading where we loaded our data using a Python library called Librosa. 

The next step involved data preprocessing. Here we prepared our data for use with the machine learning model by first removing row entries from the metadata with missing values. This preprocessing also involved the extraction of Mel-Frequency cepstral coefficients and their first and second derivatives for each cough sounds which allowed us to capture information about the shape and pattern of the cough and hence differentiate between covid-19 and healthy coughs. After this extraction, we split our data into 20% testing data and 80% training data and soon noticed that there were biases in the training dataset. To address the class imbalance in our training data we used oversampling to augment our training data.

After training our data we developed an artificial neural network using the sequential API of Keras in Python and it consisted of 5 dense layers in total each with Rectified linear unit activation functions and dropout regularisation layers to allow for a more complex network and to prevent overfitting. We trained our model using the model.fit() function and specific parameters like the training data, batch size, number of epochs, validation data and callback functions.

After evaluating our model we tested the model with a new audio file and received the model’s prediction of whether the cough was from a user with covid-19 or a healthy or symptomatic user.

In terms of the operation of our app, when our app is opened, users are given a useful disclaimer about our app before usage. After users log in, the main page is displayed where users can input their cough audio and additional information like their age, gender and other diagnostic information. After uploading this information, the model processes the input and gives the user their health status within seconds. There is also a follow-up option for users in the event that they test for covid-19. This feature allows users to connect with real medical doctors for consultations.

Respir is committed to promoting diversity, equity, and inclusivity in everything we do. Our team is made up of three Nigerian girls who come from different ethnic backgrounds and have diverse interests. We ensured to make our team up with individuals who held different perspectives towards Nigeria’s healthcare system. One of our teammates, Chioma Abone, in particular, is an aspiring medical doctor and as a result, she is highly exposed to a broad range of medical settings such as public hospitals and private hospitals. This allowed us to acquire a diverse sample population for the market research interviews we conducted. Furthermore, we prioritized inclusivity by opening our survey to a broad range of respondents from different ethnic, racial, and cultural backgrounds. This allowed us to get a good representation of our target audience’s needs.

Furthermore, regarding equity and inclusivity, we are working on implementing new features into the app to increase its accessibility such as multilingual translation and text-to-speech features to ensure that our app caters to different people from different backgrounds and does not pose any language barrier issues for our users. We are also improving accessibility by providing a low-cost physical device to serve marginalized communities.

To ensure that we become a more diverse, equitable and inclusive organization we have set specific goals.

• Inclusive workspace- We will create an environment that is safe, welcoming, and respectful of all individuals regardless of their gender or religion. We plan on doing this by providing regular training for our team on the importance of diversity, equity, and inclusion treating topics such as unconscious bias, cultural competency, and inclusivity in the workplace.
• Community outreach- We are committed to engaging with diverse communities and stakeholders to understand their unique needs and incorporate their feedback in our solution development. So far we have done this by engaging with experts, community leaders, and healthcare providers from diverse backgrounds to incorporate their feedback into our solution development.
• Accessibility: We aim to ensure that our solution is accessible and user-friendly for all individuals, including those with disabilities and limited access to technology.
• Increase the representation of underrepresented groups in our team, including women, people of colour, and individuals from marginalized communities, by setting specific targets for recruitment and tracking progress.
• Provide mentorship and professional development opportunities for underrepresented team members to support their growth and advancement within the organization.
• Conduct regular audits of our policies and practices to identify areas for improvement and ensure that we are creating a supportive and inclusive environment for all team members.

We believe that promoting diversity, equity and inclusion would foster a positive and inclusive culture which would lead to better decision-making, innovation, and productivity.

Business Model

Key resources:

• Data sets of cough sounds from covid-19 positive and negative patients to train the AI algorithm.
• Skilled AI developers and data scientists to build and refine the algorithm.
• Secure and reliable server infrastructure to handle user data and ensure data privacy and security.
• Access to funding and investments
• Marketing and outreach resources to reach potential users.

Key activities:

• Developing and refining the AI algorithm
• Collaborating with hospitals and healthcare providers for data collection
• Building and maintain the software.
• Providing customer support and feedback mechanisms to improve the app

Partners + Key Stakeholders:

• -Hospitals and healthcare providers for gathering cough sound data sets and referring potential users
• -Government health agencies for regulatory compliance and potential funding opportunities
• -Local communities for outreach and education on the app's benefits and usage
• -Technology companies for potential partnerships and collaborations

Cost structure:

• Technology development and maintenance costs
• Data acquisition and storage costs
• Marketing and outreach expenses
• Staff salaries and benefits
• Legal and regulatory compliance costs

Type of Intervention:

The app uses AI technology to analyze cough sounds for potential COVID-19 infection, providing a low-cost and non-invasive screening option for users.

Channels:

• Mobile app stores for distribution
• Social media and online marketing campaigns
• Referrals from healthcare providers and hospitals

Surplus:

• -Revenue generated from app downloads and potential partnerships with healthcare providers and technology companies.
• -Improved accuracy of the AI algorithm for detecting COVID-19, leading to more accurate diagnoses, and potentially saving lives
• -Expansion of the dataset used for training the AI, which could be used for other research purposes and potentially lead to new insights about COVID-19

Segments:

• -Hospitals and healthcare providers
• -Senior citizens and adults

Value Proposition:

Respir is an AI model that uses machine learning and deep learning techniques such as artificial neural networks to screen for the presence of Covid-19 using cough sounds. Within just a minute, Respir can screen 10-15 different cough samples making it a fast, affordable, and efficient screening method. 

Startup capital- We intend to raise our startup capital by pitching our idea to angel investors in our country who are philanthropists and are interested in bettering Nigeria’s healthcare system. We expect to receive an initial $2000 dollars in capital from angel investors and an additional $1000 from one of our teammates project sponsors, ‘Rise’ by Schmidt futures which supports teenagers ages 15-17 in developing their personal projects. Other sources of income may also come from our personal savings and the fundraisers we will host.

Revenue- As our survey suggested, target users were likely to recommend our app to their family members and friends, hence, we are likely to receive regular app usage and expect a growth of 400-500 users per month. Our revenue will also stem from not just app sales but in-app purchases and the subscription-based payment for Respir Premium. We will also generate revenue through advertising within the app.

Operating cost- Our operating costs will include the purchasing of cloud services such as amazon web services and a server hosting platform for our web app. To store our users’ data securely and efficiently we will pay for data storage, and this will depend on the size of the data, storage method and frequency of backups. To improve our AI model, continuous training is necessary and over time we will have to scale our business to cater for more data, which would lead us to hiring Artificial Intelligence engineers with expertise in machine learning. We would also have to pay security experts to ensure that our users’ data is always secure. Marketing and advertising would also take a portion of our operating costs and for legal and regulatory compliance we would need to hire and pay legal experts and consultants.

Our team leader, Emmanuela Ilok, recently won the Rise Global Challenge for her project 'CodEd' which offers a premise for different innovations. This gave her access to a competitive pool of funds of about 5 million dollars per year which can be shared amongst the 100 other Rise Global winners. These funds can be used to develop Respir which is branched out of 'CodED'.