About

zebraMD is not your physician and does not give medical advice.

It is aimed to support physicians in their clinical decision-making and cannot make decisions for them. It simply provides additional, point-of-care, supporting, and tailored information condensed from thousands of published peer-reviewed medical research articles and deidentified electronic health record (EHR) data analyses.

This is an alpha version only, for logistical testing purposes.

For patients: Ability to connect to your medical record via patient portal will soon become available. If you would like to become one of the first testers of our app, let us know here.

For healthcare professionals: Please test this app and give us any feedback on what would be helpful/what to add/what to change. You can use the feedback form button, email us, or use the highlight function to give us text-specific feedback, all of which is anonymous.

We developed a data synthesis process to create clinical recommendations based on existing literature with links to the original sources. Our output is being validated by physician specialists at UCLA and UCSF before being pushed into the live app version.

So much research and electronic health record (EHR) patient data exists all around the world already. There are even predictive algorithms created from this data, that get published in academic journals and then are never used in clinical practice — just like our original algorithm for the prediction of Acute Hepatic Porphyria. After completion and publication of the study, we thought to ourselves — what happens next? All this work for what?

So we decided to found zebraMD, a clinical AI assistant to improve diagnosis and management of rare disease.

The backbone is a platform that pulls all the rare disease data and literature together into one place, connecting it to EHR systems and patient portals, allowing us to use Gen AI for faster answers to rare disease questions, looking through millions of data sources at once, connecting the dots between papers, genome databases and individual patient records to come up with clinically actionable insights.

In addition to making our own predictive algorithms for rare disease, we also host and update other physician's predictive algorithms, adapting them to fit to raw EHR data for clinical use and improving them over time with new data that it encounters.

zebraMD is an app that can be connected to patient portals (think MyChart for example), to the EHR directly for physician use, or used with pdf medical records a user can upload for analysis. A manual symptom search is also available for a quick overview of potential differential diagnoses you might be dealing with.

The goal is to get people diagnosed, get them diagnosed earlier, and get them the appropriate care, in every single health setting anywhere in the world, no matter the zip code.

A patient should never be disadvantaged in their healthcare just because they don't have access to specialty care due to location, economic status, or long waitlists in backed up specialties.

We aim to deliver specialty care tailored to whatever department the patient is being seen in, automatically at the point of care.

Additionally, we aim to empower the patient to take control over their own health (very controversial, we know). Patients can use our app just like physicians can: They can connect their electronic medical records to our app to utilize our algorithms to hopefully find answers to their rare disease questions.

Our “OG” predictive model was originally developed for Acute Hepatic Porphyria (AHP) out of an academic research collaboration between UCLA and UCSF. As such, this model is public domain and we invite everyone to find the errors, inefficiencies and iron out the kinks! All subsequent models we develop will also be public domain.

Our article is available here, and our AHP model is available here.

We are both. Academia can be propelled forward much faster with the help of industry expertise and resources. If anyone has ever tried their hand on multicenter research projects, you know the struggle, red tape, and time-wasting that goes into that. This project necessitates as many data sources as possible in order to develop unbiased algorithms that work in any patient population in any EHR structure.

We founded a legal entity called zebraMD Inc. in 2023 in order to allow us to take on funding and collaborative partnerships with other entities, academic centers, and industries to complete this project much faster than what we otherwise could in our daytime roles as academic physicians. In fact, founding a c-corporation was a necessary step for us to get a potential integration of our algorithms into clinical practice funded, since none of the typical grant opportunities would pay for implementation.

We are now proudly supported by the National Science Foundation through their Small Business Innovation Research Program!

We are always transparent about our funding, contracts, data sources, and outputs.

The original AHP predictive model was developed with NIH research funding and industry sponsorship by Alnylam Pharmaceuticals through UCSF and UCLA. The development of the aggregated zebraMD platform and implementation journey into clinical practice is supported with industry sponsorship by Alnylam Pharmaceuticals, our existing partners since the OG project in 2020, as well as the National Science Foundation.

For development of predictive algorithms to help diagnose patients faster we currently use NSF grant funding and are pursuing disease foundation funding for each individual disease.

If you know of a suitable grant opportunity for this project, we would love to hear from you!

The data sources we use for the predictive algorithm development for AHP, hATTR, PH1, Systemic Mastocytosis and Rupture of Cerebral Aneurysms come from deidentified EHR data (classified as non-human subjects research, what that means is here) from UCLA, UCSF, Los Angeles County Department of Health, Weill Cornell and Dartmouth Health (New Hampshire).

The predictive models we are developing are freely accessible to anyone, forever.

The institutions that allow us to beta test the clinical management algorithm inside their EHR do so in the form of a quality improvement project or in the form of an academic research project with IRB review and their local Principal Investigators receive a cost payment for their time submitting and managing the IRB and data release per institution requirement. This is being paid for by the National Science Foundation.

The literature data and genomic data we extract from publicly available, published sources such as pubmed, Gnomad and Clinvar.

No profits are being generated by the use of publicly available literature and genome databases: our resulting predictive and management algorithms will forever be free to use for patients and providers alike.

On our app we also host predictive algorithms made by other researchers and physician scientists worldwide. We enable the clinical use of already existing algorithms by retrofitting them to raw EHR data of various structures and allowing for API connections to patient portals and EHR systems. If you are a scientist/researcher/physician/ organization who has made a predictive algorithm they would like to use in clinical practice, or would like to have an algorithm made with your clinical expertise, please contact us here.

We never share patient data with anyone unless the patient explicitly asks us to (i.e. sharing their data with another physician or health system for medical purposes). We use data analytics resulting from deidentified patient data in combination with existing NIH research databases to develop predictive algorithms to help get patients diagnosed and to help find treatments for patients. We only use data for projects that benefit our patients directly.

Patients have the right to withdraw their consent to participate in this research at any time. Patients can delete their account any time and all patient data will be deleted with it. More information in our privacy policy.

No patient data is stored on our platform. Clinical use of existing algorithms occurs ad-hoc when a medical record is uploaded as a pdf or a patient portal is connected via API.

The US law GINA (General Information Nondiscrimination Act) prevents discrimination by insurance companies based on your genetic test results. More information about GINA is available in our privacy policy.

Physicians: don't be afraid to order genetic testing. Every physician can order a genetic test even if they are not specialists. Medicare parts A and B cover genetic testing and genetic counseling services. Every genetic testing panel ordered through companies like Illumina and Invitae comes with free genetic counseling for the patient and the physician to use to help us interpret the findings, just like a radiologist interprets the imaging we order.

How many times do we order a head CT or chest x-ray for a patient and don't think twice about the incidentalomas we could potentially find? Yet we are prepared to deal with unexpected findings as they arise. If we need help managing the findings, we send the patient to the specialist afterwards.

Genetic testing is no different from any other part of the workup. Just because it's difficult doesn't mean we shouldn't do it.

Our project has been reviewed by the local institutional review board (IRB) at each of the participating institutions.

The IRB determination differs from site to site due to the varying availability of tools that can identify electronic health records.

IRB status:

• UCLA: IRB exempt, approved
• UCSF: Full IRB review, approved
• LA/DHS: IRB exempt, approved
• Dartmouth Health: IRB exempt, approved
• Weill Cornell: IRB pending

We cannot solely rely on grant support as it can take years (!!) to get grants even in the small business for-profit space and in a world where millions of rare disease patients continue to be undiagnosed and mismanaged every day, we do not have this kind of time.

We are currently supported by the National Science Foundation. We also continue to form partnerships with orphan drug pharmaceutical companies who develop therapies for rare and genetic diseases. Our project is mutually beneficial to our parties since it is in our best interest as physicians to get people diagnosed and then managed appropriately. Pharmaceutical companies would like to diagnose more patients and obtain more clinical (deidentified) data analysis reports detailing the efficacy and safety of their existing therapies in the community, outside of sterile, artificial clinical trials, and showing potential new therapeutic targets for diseases that currently have no treatment.

In the future, we may partner with the Centers for Medicare and Medicaid services to help us deploy our predictive algorithms nationwide to enable a faster rate of diagnosis and improved outcomes for all patients.

If you are interested in a partnership with us and would like to help us achieve our goals with zebraMD, please contact us here.

Our secret sauce is the end to end software app that can actually be used in clinical practice. It contains the development of a cloud-based platform with proprietary technology that can connect any EHR and medical records and medical databases of published research to allow for the use of retrofitted self-learning predictive and clinical management algorithms.

Our point-of-care algorithms showing predictive and management recommendations have academic references for each point to allow the reader to see exactly where the information came from. All of our algorithm output is validated by physician specialists before making it to the published app.

Using this technology, we are building the largest rare disease Co-Op in the world that actually can be used in clinical practice!