Dell SMB Series: New Ideas for Traditional Businesses
The National Institute of Cancer and Research reports an estimated 2.5 million people living with cancer in India, with over 700,000 new patients registered every year. Early diagnosis or prognosis of a cancer type is critical, but the clinician is expected to look at a huge volume of data. Imaging modalities like pathology and radiology are complicated to report, and time-consuming as well. Also, qualified pathologists & radiologists may not be available to report on the patient case immediately. Even if they are available, the number of patient cases that need to be reported daily would be significant and continues to grow.
In India, there is a significant gap between the number of pathologists available to report and the number of samples being collected, leading to a backlog at the start of each day. This is a big disservice, given the top quality of cancer care available in India. There’s no cure for cancer, but fortunately we have algorithms to help treat it earlier.
This is where Onward Health comes into the picture. Having worked together previously in the patient engagement space, founders Dinesh Koka and Vikas Ramachandra had multiple conversations with oncologists before they concluded that most health scoring tools were either non-existent for cancer or had been built for the western population. They decided to focus on using machine learning platform and with the help of predictive analytics started Onward Health to try and provide value to clinicians.
Onward Health uses machine learning expertise to build a portfolio of diagnostic tools in the form of classifiers and analytical tools. They establish clinical partnerships with diagnostic labs based on the match between the tools and the clinical workflow of the diagnostics provider. These labs bring in clinical expertise, exposure to real-world challenges, as well as validate the tools. In most cases, there needs to be 1-2 areas where the tools they build will be of significant value to the partner, in terms of productivity or for extending the clinical service beyond an existing network. These partnerships are usually long-term because the value from leveraging sophisticated techniques in machine learning will be realized over time, because of the collaboration.
They also work with leading pathologists, radiologists as well as oncologists to build machine learning (ML) models. The clinicians look at the base diagnostic and scoring tools and help identify the areas in which they can leverage the tools. They also work together in identifying additional features that could be added to the tool to improve its utility and effectiveness.
The Onward assist platform will be provided on a subscription model, where the centres pay a subscription fee to use the platform. These tools will be deployed in a setting where the reporting doctor can access the analytical tool within their workflow. Additionally, for certain partners, other commercial models are explored based on a specific need.
Onward Health provides tools in computational pathology and mammography, leveraging computer vision techniques and ML algorithms.
In pathology, the imaging algorithm is designed to detect different types of cells. Once detected, the algorithms also process each cell to determine various properties such as shape, grade, texture, colour distribution features, as well as distance to the tumour and other spatial neighbours. These features are then used to train machine learning (ML) models. Expert pathologists go through the results and validate and provide their inputs.
The mammography diagnostic tool, on the other hand, evaluates a mammography image and helps classify the cancer cells as positive or negative. It identifies the tumour region and tries to find similar regions on the other views through image comparison. Then, for each of the identified suspicious regions, image-based features are extracted. Expert radiologists go through the results, validate and confirm the analysis through biopsy results.
The Onward Assist platform is all set to provide a variety of benefits to diagnostic labs all over India.
It significantly increases productivity by 30-40%, which is reflected by the reduction in the time taken to complete the reporting of a slide.
Onward Health works in two modalities in the areas of breast and cervical cancer- histopathology and radiology (mammography), as the oncologists treating these cancers are looking for diagnostic and prognostic information. Three key factors differentiate them from others in the market- domain knowledge, technical capability and strong clinical partnerships.
Domain knowledge: The founders bring to the table a combined 30 years of experience in the healthcare industry across various streams of product management, sales, provider management, AI/ ML, and analytics experience.
Technical skills: Based in the CIE Incubator of IIIT Hyderabad, they primarily leverage the academic connect, especially in the areas of Machine Learning & Artificial Intelligence. They grasp the expertise available both to validate their work as well as discover the latest technologies in this field.
Clinical partnerships: They have partnered with leading providers in the space of cancer care and diagnostics services. These include 2 out of the top 3 cancer institutes in India, and pan-India pathology services network with a presence in 15 states.
Cancers can be detected early on and treated, before they have a chance to grow and spread. In addition to mammography and histopathology tools, Onward Health is also building tools in the space of 'virtual staining' which will evaluate an initial stain (H&E) to decide on the type of follow-up special stain to be done.
Through the strong clinical partnerships, they look forward to tapping into clinical expertise and at the same time build relevant solutions based on real-time clinical workflow.