Gradient Health, Inc: Equitable Healthcare A.I.

Our Organization

Gradient Health, Inc

What is the name of your solution?

Gradient Health, Inc: Equitable Healthcare A.I.

Provide a one-line summary of your solution.

Gradient is building the foundation of healthcare A.I. by labeling the world’s pathologies.

What specific problem are you solving?

A global, foundational problem that plagues healthcare AI development is locating access to large, diverse medical datasets and accessing datasets from developing nations around the world, specifically in the Global South. Majority of data used in A.I. applications come from wealthy nations, ultimately affecting the world’s most vulnerable populations. Most models currently use data from small, insular datasets, and even less use data from a geographically diverse dataset. In addition to limited access to medical data, indexing huge amounts of medical data without exposing the personal data of patients is difficult and limits the imagination of A.I. researchers and developers. A different problem exists for patients: while imaging volumes are increasing worldwide, the expertise to interpret them is not keeping up. This problem is especially acute in developing countries, where certain regions of the world have access to less than 1 radiologist per 1000 people. This anticipated radiologist shortage could lead to worsening of physician burnout, delays in patient care, and ultimately poor clinical outcomes. 

What is your solution?

Gradient Health, Inc is targeting a foundational problem in the development of healthcare A.I. - large, expertly labeled datasets from a geographically diverse base. We aim to index a broad swath of imaging pathologies, in both labeled and raw formats, from around the world. We seek to improve the index by including marginalized populations, with a focus in Africa and South America. We want to make this database, which will update in real-time, available to any researcher or AI engineer as they seek to develop algorithms that can be deployed in low-resource settings. Gradient does this by practicing non-exploitative data practices in developing countries in the Global South by providing picture archival and communication (PACS) system and remote radiologist training for hospitals in developing nations.   

Gradient Health will partner with hospitals and imaging centers in multiple countries across the world with special attention to countries residing in the Global South. Our technology is vendor-neutral and uses advanced natural language processing techniques to organize the images according to radiology findings. To ensure that low resource settings are not left out of the loop, we believe that it is imperative that they are also able to take part in the monetary gains arising from their datasets. To ensure this sustainability, we plan to adopt a revenue share model with our data sources that is fair and comparable to that which would be achieved in developed countries. 

We seek to not only index the world’s imaging pathologies, but to grow our network of radiologists to label them. Doing so will not only dramatically scale up the development of computer vision algorithms, but also bring additional income to radiologists in South America and Africa. With our database, it would be possible to generate a large cohort of patients based on their imaging pathologies and subsequently develop algorithms that could one day be fairly and accurately deployed for the very people that they were trained on. 

Imaging data is most useful to research and technology development if there is associated context, such as patient clinical presentation and imaging findings. Gradient Health is eager to be a part of a global effort to break down the barriers to access and curation of data and most importantly, to fairness in data usage. The following is a brief description of our accomplishments so far:

1. Indexing and dashboard software. Most academic or imaging centers, whether in the US or abroad, are not even aware of the data sitting in their servers.  We designed a dashboard based on radiolong “findings,” a collection of terms and descriptions used by radiologists to describe the pathologies in the images. The dashboard will also reflect patient age, demographics, and geographical location, among other data. Our software can be integrated into any vendor’s PACS and reporting system. The software will update in real-time and will allow for immediate creation of patient cohorts based on their radiology findings. 

2. De-identification software. We continue to refine our de-identification software. Although we anticipate that the majority of DICOM images will already be de-identified, we want to add an additional layer of privacy by detecting and redacting burns out protected health information. 

3. Labeling software. We have developed web-based image labeling software that is not only user-friendly, but encompasses the majority of tasks that machine learning engineers could want, including bounding box and segmentation. A major benefit of our software is that multiple users can label images remotely, without the need to download additional software. In addition, the newly labeled data can be transmitted seamlessly and securely to the development team. In addition, our labeling software can also function as a PACS. Although it lacks some of the post-processing functionality seen with established companies, our software will still represent a significant step-up for certain imaging facilities in Africa.

We seek to build on top of the resources we built in-house and ask for funding to do the following:

1. Further refine and subsequently integrate our software with current and potential radiology partners in Africa and South America. 

2. Develop a centralized database and dashboard of radiology findings that can be searchable through our online portal. We will make this index available to the public. Once a user generates a cohort of patients with the desired imaging findings, they will be able to ping the data sources and initiate either a data transfer or a labeling contract. A data use agreement will be implemented, and one of the terms will be a requirement to make any commercialized software available to their data sources (ie. hospital or imaging center) at below-market cost. Publications stemming from this data network will also need to reference the data source. 

3. Work with our network partners, whether in academia or industry, to leverage our software tools and index of imaging pathologies to build algorithms that can meaningfully impact underserved populations. Examples include: intracranial bleeds, pneumothoraces, malpositioned feeding or endotracheal tubes, viral pneumonia, such as that seen in COVID-19, presence of abscesses that can be drained percutaneously, etc. 

Who does your solution serve, and in what ways will the solution impact their lives?

This solution solves A.I. companies and academic researchers’ needs for large, diverse annotated medical images. This project also serves hospital systems by permitting them to have an active role in the development of A.I., improves their understanding of how AI models are deployed, provides them a PACS system and educational training for radiologists. There are also employment opportunities for trained radiologists to label data for Gradient Health, Inc. The data sourced from these communities will be anonymized and labeled for use in training and validation of AI algorithms that could then be deployed for the very people whose data was accessed. This will hopefully improve the speed and quantity of healthcare delivery while providing a revenue share model that is fair and respectful. The revenue share model is necessary to incentivize hospitals to engage with the data network. Because images are obtained in a DICOM standard, these images can also be deployed in developed countries, but at a higher price point, which would incentivize well-funded industry partners to participate as well.

How are you and your team well-positioned to deliver this solution?

We have a diverse, accomplished group of people working to solve this problem. Our founders, Josh Miller and Ouwen Huang have previously worked together to a computer vision agricultural technology company FarmShots, a satellite imagery platform that was deployed across Africa and Brazil. Our Chief Medical Officer, Dr. Sophie Chheang is Interventional Radiologist and Assistant Director of Informatics at Yale. We are currently partnered with Telelaudo, a teleradiology company that provides remote radiology reports to imaging centers and hospitals in Brazil and other Portuguese-speaking countries. Nico Addai is a Research Consultant for MIT STEP lab and alum of MIT's Data+Feminism lab where she is trained in A.I. ethics development. 

Which dimension of the Challenge does your solution most closely address?

Build fundamental, resilient, and people-centered health infrastructure that makes essential services, equipment, and medicines more accessible and affordable for communities that are currently underserved;

Where our solution team is headquartered or located:

Our solution's stage of development:

How many people does your solution currently serve?

20

Why are you applying to Solve?

Gradient is applying to Solve because we recognize the scale of the problem and our proposed solution. Artificial intelligence and developments in healthcare technology will affect the whole world, and we believe that partnering with MIT Solve will allow us to have greater impact with experts in the field of ethical design and development.

In which of the following areas do you most need partners or support?

Financial (e.g. improving accounting practices, pitching to investors)

Who is the Team Lead for your solution?

Joshua Miller