Investment Notes - Arintra
At TEN13, we are continually focused on partnering with teams building products that are uniquely solving critical problems for their customers within large addressable markets.
We are therefore excited to announce our recent investment in a company solving one of the biggest problems in the US healthcare system.
We are delighted to be backing a world-leading team of technical founders with PhDs in machine learning and computer science who are building Arintra, an autonomous medical coding platform.
Arintra leverages the latest advancements in AI to solve a more than US$100 billion annual problem tied to revenue claiming for US hospital systems. With Arintra, customers can materially improve revenues and cash flow.
Problem and market opportunity
US hospitals primarily generate revenue via claims for reimbursement for treatments they provide. These claims are submitted to large health insurance companies or directly to the Federal Government (covered by Medicare). This is big business - in 2021, US healthcare spend grew to over $4 trillion, accounting for ~18% of US GDP.
So where does medical coding fit into the US healthcare system and what is it? Medical coding is core to this claim process and has a first-order effect on revenue generation for hospitals.
Whenever a patient receives care in the US in a hospital, the services provided, the procedures performed, and the diagnosis received are recorded in an Electronic Health Record (EHR) software system. These records serve as a digital version of the patient's clinical chart or record. Next, the charts are “coded”, meaning that they are assigned a specific combination of codes that tie back to the exact treatment that the patient received. These codes are then sent to the billing department of the hospital to generate the claim to be sent to the insurance party and bill for revenue.
However, this coding process - central to how a hospital generates revenue - is still primarily completed by humans, which is time-consuming, complex, and prone to errors. Compounding these challenges, there simply aren't enough human medical coders to cover the backlog - studies show there is actually a 30% shortage. These forces combined have a material impact on hospital cash flows, with sometimes millions of dollars of uncoded claims outstanding each month for hospitals.
Vision & Solution:
Arintra uses proprietary AI models, built on top of open-sourced large language models (LLMs), that fully automate the majority of the generation of these codes, with no human intervention.
Arintra materially improves revenue and cash flow for hospital systems - versus the current human solution, it is significantly faster, more economical, and more accurate.
Importantly, for Arintra’s customers, there is no change to the existing workflow of patients, doctors, medical coders, or insurance companies. This enables seamless integration of Arintra’s solution into the workflows of all stakeholders. Automating medical coding through AI makes the whole process more efficient, accurate, and reliable
This is not just “AI hype”, but a real application leveraging the advances in AI technology to solve a very large industry-wide and priority problem.
Why we invested:
Strong technical team, with a deep understanding of the problem:
- The company's founders, Nitesh and Preeti, completed their PhDs at the University of Maryland in computer vision and computer science, respectively. Having moved from India to the US, they experienced the inefficiencies of medical coding as patients themselves - after several visits to the hospital, they were stuck with tens of thousands of dollars in expenses not being reimbursed after their treatments were either inaccurately coded, or there were long delays of codes being submitted to the insurer.
- Over the last 2 years, this exceptional founding team has gone on to build a unique platform, with its own state-of-the-art AI algorithms built on top of the latest open-source models.
Real AI use case and already showing signs of material long-term opportunity:
- We believe that vertical specific proprietary AI models, such as Arintra, built on top of open-source base large language models, being trained on and leveraging unique proprietary datasets is where real value is likely to sustainably accrue in this new AI landscape.
- The results of their automated coding platform are already significant; with material improvements to accuracy and speed, for large hospital systems this leads to reduced claim denials, less under-coding (more revenue for hospitals), and reducing cashflow cycles from receivables due to quicker claims.
Addressing a ~$100+ billion problem for one of the largest sectors in the US:
- This is a very large market opportunity for a vertical AI-powered software play. US National Health Expenditure grew to ~$4.3 trillion in 2021, with ~5 billion claims per year.
- Inefficiencies in medical coding, with denied insurance claims, fines, and missed revenue, are costing US hospitals over $100 billion annually.
CEO and Founder of Arintra, Nitesh, with TEN13’s Seamus and Peak XV’s Pushpak, in San Francisco
Investing alongside our friends at Peak XV (previously Sequoia India / SEA), we are excited to partner with Nitesh, Preeti and the team, who are changing the game for US healthcare.
Safe Travels!
Seamus and the TEN13 Team