Chestify AI

We are solving 3 primary challenges in medical imaging today;

1. The inadequate radiology infrastructure and fewer radiologists in Africa and The Caribbean starting from Ghana.

    Ghana has about 70 certified Radiologists, Liberia has 2 as of January 2022. In the majority of sub-Saharan African countries, there is a marked paucity of diagnostic equipment available to the clinicians.

2. Inefficiencies and high inaccuracies in imaging diagnostics amongst radiologists and clinicians.This has led to misdiagnosis of medical images and loss of lives due to wrong medication given out to patients who have either been diagnosed         with false positives or false negatives.

    Also, The number of medical images keeps skyrocketing while the number of radiologists stays relatively the same.

3. High medical cost in imaging diagnostics. $750 Billion is wasted annually in the US  on diagnostics errors and unnecessary testing. 

According to the Ghana Statistics Agency, Ghana loses about  $5.9 million on diagnostic errors annually.

- We have developed deep-learning algorithms for diagnosing medical image pathologies;

a. Our product increases the diagnosis accuracy of the clinician and urgent patients can be diagnosed at a 10 times faster rate than the traditional way.

Example: Globally, about 2.5 million people die from pneumonia. A report from UNICEF states that over 700,000 children die annually from pneumonia, that is 2,000 deaths daily, 1 death every 43 seconds. And almost all of these deaths are preventable. Our algorithm for Pneumonia has a 95% accuracy rate.

b. Our product optimizes radiology workflow, reduces operational drudgery and medical cost by 25%.

Health facilities and medical practitioners who use our platform have a 90 - 95% enhanced rate of accuracy in their workflow, fast diagnostic reporting procedures with a 10 times increased productivity and evade operational drudgery and complexities using the platform - (user-friendly interface).

Our target market includes Hospitals, Lab test centers, Private Radiologists.

Hospitals and Lab Test Centers do not have adequate radiology infrastructure and the clinicians that diagnose pathologies on x-ray images have an accuracy rate of 40%  at the most and the least they do not know what to diagnose.

The total number of Radiologists shared amongst the Ghanaian population would be 400,000 Ghanaians to 1 Radiologist, hence the pressure on the radiologists decreases their efficiency rates and are unable to diagnose appropriately.

This reality affects patients the most.

Patients who are wrongly diagnosed are made to take medications for diseases they do not have or vice versa.

Our solution serves as an assistive aid which optimizes medical workflow and improve diagnostic accuracy to about 90%. It also has an interventional radiology functionality that gives analysis on images if there is a probable disease that might jeopardize a patient's health and well-being hence clinicians are able to recommend good eating and exercising habits and prescribe medication to totally eliminate the infant disease from developing.

We understand the need for collaboration to accelerate our solution's impact in our targeted communities. We work with radiologists, radiographers and healthcare providers who have expertise in the radiology and medical imaging industry.

Our team is made up of devoted young individuals from diverse backgrounds in Software/Machine Learning Engineering, Business Development & Strategy and Radiology.

CEO & Team Lead Mustapha Zaidan has over 8 years of software engineering and has aided in the development and growth of 15 software products in Ghana including Dvash Platform which seeks to address the challenge of food insecurity and food waste through blockchain technology.

Michael Osei-Gyeabour, Co-Founder and CBO has 3 years of software development, 2 years in Cybersecurity, 6 years of financial modeling and Business development and experience in Industrial engineering where he worked with Diageo. Francis Collins, a certified Radiologist in Ghana, Paa-Kofi Anderson, a Full Stack Software Engineer, Ange Mylene, a research analyst with a background in probability & statistics.

Our advisors include, Dr. Augustina Peprah who is currently the president for the Ghana Radiologists Association, Nick Barba - Senior StartUp Manager at Future Founders and Wendy Lea who is our strategic advisor.

We currently have 20 Hospitals, and 65 Clinicians (approximately 32 Doctors) using the product.

We have undertaken 75 successful pilot tests, out of which 48 organizations have shown interest in paying to use our platform.

  In our Roadmap, our product is able to diagnose 14 pathologies on X-ray images including Pneumonia, Cardiomegalia, Atelectasis, Edema and others as listed in our pathology documentation.

We aim to fully build algorithms that are able to detect 20 pathologies by the end of 2023 and 135 pathologies by Q4 of 2024. As we collate data from our respective healthcare partners, our radiologists annotate the datasets making sure we feed our system with accurate data.

We are in the third phase of getting a Ghana FDA license as a medical product and we estimate to get the license by Q3 of 2023, After which we would begin selling our product to clients in order to start generating revenue.

In humanitarian contexts, our AI platform enables emergency preparedness and prevention by providing healthcare settings with reliable and timely diagnosis of X-Ray images. This allows for early intervention and better patient outcomes. Furthermore, in the event of an emergency, our platform can provide healthcare workers with timely and accurate diagnosis data, which can help inform the deployment of resources, and lead to better-informed decisions and improved response times. Finally, the data provided by this platform can help in the recovery process, by assisting in the formulation of strategies and plans that strengthen a particular region's ability to cope with future crises.

As a startup focused on addressing a critical healthcare challenge in Africa and the Caribbean, we recognize the importance of leveraging partnerships to achieve our mission. When we are selected for the program, we would be provided access to partners who can provide valuable expertise, resources, and connections that can help us scale and grow our impact. 

We believe that participating in this program will help us build a strong network of supporters who can provide us with the financial, technical, and strategic resources we need to achieve our vision. 

Additionally, this program can also provide us with the exposure and visibility we need to attract new investors, collaborators and strategic partners who are aligned with our mission. Ultimately, we believe that building strong partnerships is critical to the success of our startup and the program provides an opportunity to connect with the right partners to help us achieve our goals.

In our current stage, We are more keen on getting into partnership with  medical research facilities that specialize in Radiology and biomedical imaging . This, we believe, would be a good way to get consistent and accurate annotated data-sets for further development of our ML algorithms and would further substantiate our authenticity in the radiology industry. 

We are certain that MIT SOLVE would have partners in medical imaging research and access to other relevant industry players in the radiology field.

Chestify AI platform diagnoses pathologies on X-Ray images and It is unique in that it benefits both medical practitioners and patients alike. For practitioners, the platform saves them considerable time and effort by quickly and accurately analyzing X-Ray images and providing them with a comprehensive list of possible pathologies. For patients, the AI platform ensures that their pathologies are diagnosed in an efficient and reliable manner.

In humanitarian contexts, our platform enables emergency preparedness and prevention by providing healthcare settings with reliable and timely diagnosis of X-Ray images. This allows for early intervention and better patient outcomes. Furthermore, in the event of an emergency, our platform can provide healthcare workers with timely and accurate diagnosis data, which can help inform the deployment of resources, and lead to better-informed decisions and improved response times. Finally, the data provided by this platform can help in the recovery process, by assisting in the formulation of strategies and plans that strengthen a particular region's ability to cope with future crises.

As an AI-powered healthtech startup, our solution is innovative in several ways:

1. Accuracy: Our machine learning algorithms use machine/deep learning techniques to analyze medical images precisely and efficiently, increasing the accuracy of diagnosis and minimizing human error.

2. Speed: Our algorithms can analyze numerous medical images within seconds, saving time and reducing the wait times for patients.

3. Accessibility (Teleradiology): Our web-based platform allows physicians and radiologists to interpret medical images remotely, which enables medical professionals from anywhere globally to offer diagnoses without having to be physically present.

4. Cost-saving: By leveraging AI-based tools, we are creating efficient and automated approaches to patient care delivery, which is reducing operational costs by over 30% and boosting efficiency.

5. Interoperability: Our platform is easily adaptable and can integrate with other healthcare AI tools.

Overall, our AI-powered platform helps bridge the gap between excess demand for radiology services and the limited number of radiologists, ultimately leading to better patient outcomes.

As a healthtech startup, We aim to fully build algorithms that are able to detect 20 pathologies by the end of 2023 and 135 pathologies by Q4 of 2024. As we collate data from our respective healthcare partners, our radiologists annotate the datasets making sure we feed our system with accurate data.

Using machine learning algorithms to diagnose pathologies on medical images, our impact goals for this year and the next year includes:

1. Improving the accuracy of our machine learning algorithms for diagnosing common pathologies by at least 20%.

2. Expanding our reach to new medical facilities and partnering with more doctors and hospitals to provide more access to your technology.

3. Conducting research and publishing papers to demonstrate the efficacy of our technology in improving patient outcomes and reducing healthcare costs.

4. Engaging with government and regulatory bodies to ensure our technology remains compliant and accessible to patients regardless of socioeconomic status.

5. Implementing continuous professional development (CPD) programs to upskill and certify our core team members, radiographers and other medical staff to increase their knowledge and proficiency with our technology.

For the next five years, our impact goals might include:

1. Scaling up the technology and expanding to international markets, specifically in Africa & the Caribbean.

2. Developing and customizing different versions of the technology for specific imaging applications, such as mammography, ultrasound,fluoroscopy, PET etc.

3. Developing standards and guidelines in partnership with medical societies and regulatory bodies to ensure that our technology is being used ethically and safely.

4. Engaging with and empowering patients to be more proactive in managing their health outcomes and decision-making via our platform.

5. Establishing research grants and fellowships to attract top talent to our organization and catalyze scientific discoveries in the field of radiology.

To achieve these impact goals, we are focused on:

1. Continuous improvements to the accuracy and robustness of your algorithms, incorporating feedback from users and benchmarking against industry standards.

2. Leveraging marketing and communications strategies to build brand awareness and increase uptake amongst our target audiences.

3. Navigating regulatory requirements, licensing agreements, and reimbursement models in different markets to ensure accessibility to our technology and drive sustainable growth.

4. Building strategic partnerships with healthcare providers, radiology societies, patient organizations, and government bodies to create a network of advocates and influencers who can help amplify our impact.

5. Empowering and investing in our core team of experts, technologists, and clinicians to drive innovation and foster a culture of learning and growth.

Our current KPI's includes:

1. The number of health facilities and medical practitioners signed up unto our platform.

2. The number of low-setting underserved communities we have implemented our solution in.

2. The number of patients that are diagnosed with our platform and receive accurate medical care.

3. Volume of data-sets we are able to churn weekly from various health facilities.

4. The readability accuracy rates of our models in diagnosing the pathologies on biomedical images.

5. Number of medical experts, radiologists and product development experts we have on the team.

7. Turnaround time in using the product.

8. User Satisfaction by health facilities, medical professionals and patients.

We utilize machine learning algorithms and artificial intelligence technologies including;

• Convolutional Neural Network :Convolutional Neural Networks (CNNs) have become an important tool in machine learning for radiology diagnosis. CNNs excel at recognizing patterns in images and have shown promising results in the analysis of medical images such as CT and MRI scans.
  In our radiology diagnosis, CNN is used to detect abnormalities or predict the likelihood of certain diseases based on medical imaging data. The CNN is trained on a dataset of labeled medical images, and the algorithm learns to recognize patterns and features in the images that are indicative of various conditions.

• Out-of-Distribution Detection : Utilization cases of an out-of-distribution detection method employs a reconstruction loss Only images which are similar to those within the training distribution are allowed to be processed.
• Pathology Prediction : We used a DenseNet-121 architecture which is shown to work well on chest X-rays. It is trained with the train-validation-test split as in the ratio (70%, 10%, 20%).
• Prediction Explanation : We used the gradient saliency map which has precision and gives relevant information on predictions and accelerates the clarification of why a network made a prediction.
• Web Computation : We created a pipeline using Open Neural Network Exchange (ONNX) to rework models trained in the PyTorch library.
• AWS SageMaker for building machine learning models.

In our Social Impact Policy Statement, we have the following tenets that addresses diversity, equity and inclusion;

1. Establishing  partnerships with community health organizations and social service providers to develop outreach strategies to target vulnerable and marginalized populations in the community. 

2. Implementing programs and services that consider language barriers, cultural differences, and lack of access to medical care among  populations in underserved communities we target.. 

3. Exploring technology-driven initiatives that make radiology diagnoses more accessible, such as online scheduling, remote consultations, and cheaper home delivery of radiology services(teleradiology). 

4. Investing resources in training and education of clinicians to ensure team members treat every patient with respect and understanding of their individual needs.

5. Offering  services at a lower cost to vulnerable populations, either entirely or through special discount offers. In our pricing models, we provide payment plans, lower fees for uninsured patients, and discounts for those utilizing the nation's healthcare programs. 

6. Evaluating procedures and policies to identify possible areas of discrimination, and address them in order to promote inclusivity. A typical example is how gay individuals are treated with disdain in most health facilities in Ghana and most African Countries.

7. Collecting feedback from patients and stakeholders regularly to assess quality of care and ensure access to quality radiology services for vulnerable and marginalized populations. 

8. Leveraging data to determine areas of need, develop strategies to improve outreach, and identify underserved areas to be addressed. 

By implementing these strategies, we are  promoting inclusivity, and ensuring  that marginalized and vulnerable populations have equal access to quality healthcare services.

As a healthtech startup, Our revenue model comprises 3 main pricing strategies(projections);

1. Yearly licensing : $8500.00 base price and given by quotes to larger organizations. 

2. Monthly Subscription : $1850.00

3. Pay per Use credit system(per image) : $1.00 

The license model is for Healthcare organizations  who would serve as a partner vendor to distribute the tool to their member hospitals.

In addition, we are exploring partnerships with medical imaging equipment manufacturers to develop integrated systems where our algorithms can be used to optimize imaging analysis and reduce costs associated with manual interpretations. 

Another potential revenue stream we are exploring is to develop partnerships with pharmaceutical companies and medical device companies to provide early-stage disease detection using our algorithms.

Also, We intend to fund our project through external investments and grants, and collaborations with established companies and other research partners. 

We believe that by combining our innovative approach with a sound financial strategy, we can make a significant impact in the healthcare industry and become financially sustainable over time.

As an AI-powered healthtech startup, there are a few ways we are working towards financial sustainability:

1. Subscription-based model: We offer our services to hospitals and clinics on a subscription basis, where they pay a monthly or yearly fee for the use of our technology. This model has been successful so far, as we have seen an increase in the number of healthcare facilities using our services.

2. Partnership with healthcare providers: We have partnered with healthcare providers to integrate our technology into their existing workflows. This approach allows us to reach a wider audience and generate more revenue.

3. Grants and funding: We have received a few grants and funding from organizations that support healthcare innovation for our research and development. This has helped us grow our technology without taking on too much debt. The organizations include: UNICEF StartUp Lab & Afritech StartUp Bootcamp.

4. Cost optimization & Country Price Indexing: We continuously analyze our costs to ensure that we are operating efficiently and that our product pricing matches the country pricing index in the radiology and radiography industry and economic climate. This helps us maintain a healthy profit margin and reinvest in our business for growth.

Overall, we believe that our approach to financial sustainability is working well, and we are committed to continuing to grow our business while also ensuring we provide high-quality services to our clients and patients.