Sepsis Watch

Sepsis affects humans globally. In the US alone, there are 1.7 million cases and 270k deaths annually as per AHRQ's most recent analysis of US health data. At $43bn annual cost (as per AHRQ), Sepsis is the single largest expense in US healthcare.

Sepsis is a condition in which the body responds improperly to an infection. The infection-fighting processes turn on the body causing organ failure or death. Sepsis is imminently treatable with common available medication if caught early. However, complexity involved in determining onset of sepsis  introduces significant variability and delay since it currently largely relies on (human) clinical judgment.

We provide consistent early detection of sepsis so that timely treatment can be initiated to save healthcare costs and human lives.

Developing a robust AI solution in healthcare requires access to rich curated patient data. 

We licensed clinical AI, Sepsis Watch, developed by Duke University using patient data from the Duke health system. Sepsis Watch, live in Duke Health system since 2018, analyzes in real-time the vitals, laboratory results, medications and other parameters to predict his or her risk of sepsis during hospital stay. 

We got a grant from NIH to prove that a system developed and running successfully in Duke could solve sepsis elsewhere. As part of this grant, we showed that Sepsis Watch would work outside Duke and the first pilot of Sepsis Watch will go live in an Ohio based health system in November 2023.

Sepsis Watch facilitates timely and consistent treatment at scale in a hospital. The platform will consist of an Electronic Health Record (EHR) data ingestion engine that will feed the AI model. A user interface will enable clinical teams to get alerts and then track treatments for sepsis.

The user interface is a key step to ensure that clinicians/customers engage with the AI models that have been running in the Duke Health system since 2018. It has already had a significant impact at Duke by improving their clinical quality scores and operations, which also positively impact their financials.

Sepsis Watch will be used in acute-care hospitals by clinical staff that are currently under severe stress from pressure to address sepsis. It will directly save patients lives who are otherwise at the mercy of clinical variability in hospitals.

The co-founders, Srikanth Muthya and Ajit Thomas, have been actively working with Duke University as part the NIH grant since January 2021 to show that Sepsis Watch can save lives in hospitals.

In addition, Srikanth has over 25 years of experience in health care informatics. He held senior level positions at GE Healthcare and Philips Healthcare. He also developed remote monitoring solutions for developing countries at his first start-up. Thus, he brings proven healthcare informatics and interoperability experience not just in large corporations, but, also in a startup.

Ajit has experience building product, sales, brand and customer engagement strategies for a start-up that he grew before it was acquired by an Insight Partners & Genpact backed company (Enverus aka Drilling Info). At Cohere-Med, he also led the commercial and scientific grant application process that helped secure the non-dilutive NIH STTR grant of over $1m (for Sepsis Watch). Since he joined Cohere-Med, he led all discussions with the clinical leadership at the Ohio based health system to help us obtain our commercial traction. He oversaw development of the medical dictionary for medications for Sepsis Watch.

The AI model was licensed from Duke University where it is running in routine clinical setting since 2018. It was developed by a multi-disciplinary team coming from the schools of medicine, computer science engineering and biostatistics with difficult to obtain patient data (which came from their own health system). Getting (1) access to highly curated medical data to (2) develop a model with multidisciplinary skills for the (3) model to be used in routine clinical setting since 2018 makes our AI model unique.

The NIH grant has allowed us to show that the model can be applicable outside Duke and can be improved even further with broader access to data as it gets installed in other health systems.

Our solution addresses sepsis- a disease which reduces quality of life among survivors by affecting organ function or causes outright death. This is a global issue and our solution can be used globally in hospitals.

Sepsis Watch AI model uses Recurrent Neutral Network approach to conduct longitudinal analysis using real-time data of the patient to assess their risk of becoming septic during hospital stay.

The model was developed using highly curated patient dataset from Duke Health system.

The model will use federated learning approach to further fine-tune as it gets installed to address sepsis in health systems outside Duke. This captive access to patient data gives access to difficult to get patient data.

We conduct regular studies to ensure that the model performance is consistent across locations, sex and race. This ensures that the health systems can place high reliance on using our product irrespective of the patients' sex and race.

We install our system inside the health system's firewall in their secure servers. This ensures that all patient data always remains inside the health systems control.

Our goal is to reduce costs and deaths in US healthcare by first addressing problems in US hospitals with sepsis.

Within the first year, after our solution enters routine clinical setting at the Ohio based health system, we will be ready to scale. The next one year would be building the components required for scale.

Since we will have access to patient data, within five years, we will extend our solutions to other diseases such as cardiac decompensation (for which we already have a license from Duke) and Advanced Kidney disease.

We have three FTE, one part-time employee and one contractor.

The founding team of Srikanth and Ajit have been together since November 2020. The current team, with employees and contractor, has been in place since February 2023.

Given that we are a small team of five individuals at this early stage, we haven't formulated a formal DEI policy. We look forward to receiving guidance from experienced mentors on this regard.

Srikanth is the head the technology and works closely with Ajit on product strategy. Duke University also provides guidance on a weekly basis as we install the product in Ohio.

We have one FTE who works on developing the software infrastructure under Srikanth's guidance. We have one part time data analyst who is developing python scripts for real-time data curation. 

Our one contractor provides expertise on extraction of data from Epic, an EHR system used in Ohio.

We have partnered with Rhapsody to develop the data-extraction engine to feed our AI model inside the health system.

Duke University, as advisors, also help us reach out to other health systems interested in solving sepsis immediately giving us visibility to a larger sales pipeline.

We have engaged one clinical leader, who was President at a hospital, to provide us operational insights of a hospital as we get close to commercial installation of our product.

We are seeking industry experts who can help with the commercial scaling (in acute care specifically) right after we go live in Ohio.

We are also exploring strategic partnerships with vendors to health systems who can improve our ability to scale to health systems.

We currently have a NIH grant that funds our immediate development of the product as it gets ready to go live. We are exploring other government grants (NIH, Arpa-H) that provide non-dilutive funding. These funds cannot be used for commercial operations.

Our business model is B2B SaaS. We expect our first subscription revenue out of Ohio next year. 

We have raised $265k from angels. We plan to raise a seed round later next year for commercial scaling.

Our current burn is ~$40k a month. This is being funded by NIH.

Soon after we go live, this is projected to increase to $70k a month as we scale up our commercial operations to hire one additional engineer and one sales person and for marketing and travel for sales.

We request $100k.

$80k will be used to hire sales person (that does not get funded by NIH grant). 

The remaining amount will be used for marketing ($10k) and domestic sales travel ($10k) by the sales person to health systems that can be future prospects.

While Srikanth and Ajit are experienced managers, this will be the first time they will be running a firm that will seek venture capital. As such, we look forward to Cure residency to give us the pathway to raise institutional capital for a solution that can save hundreds or thousands of lives and billions of dollars.

We also seek advice on initiatives like DEI from seasoned professionals who have experience on these matters as we look to build our team next year.