The solutions help pathologists and radiologists arrive at a quicker and accurate conclusion to the diagnosis. This is turn helps oncologists decide on the appropriate cancer treatment.
The advancement of healthcare needs to solve for provision of faster and accurate treatment. To predict the right outcome, doctors rely on technology for decision making. The right data fed into the system and DeepTech intervening at the right juncture in the diagnosis to treatment cycle can work wonders while treating diseases that is a literal race against time.
Onward Assist is a cancer analytics platform providing automated diagnostic tools. The solutions help pathologists and radiologists arrive at a quicker and accurate conclusion to the diagnosis. This is turn helps oncologists decide on the appropriate cancer therapy/treatment.
Quicker diagnosis and accurate treatment with AI:
Founders Dinesh Koka [CEO] and Vikas Ramachandra [CTO] previously worked with healthcare providers and cancer centers. In conversation with oncologists, both Koka and Ramachandra, the former reported the success rate of a treatment decision is half or less than that. Complicating the above scenario, oncologists have to analyze a huge volume and variety of data to arrive at a specific conclusion. Koka says this problem statement could be attempted with artificial intelligence and machine learning.
Role of DeepTech:
In cancer treatment, the study of tissues is important for the right diagnosis of the progression and behavior of the disease. In cycle a patient may first get screened asymptomatically or symptomatically. A radiologist and pathologists study the biopsy slides and Xray, CT scanners and MRI images and provide inputs on the diagnosis to the oncologist. The latter then stacks the information together to decide on the treatment taking into account the prediction of patient’s response to it. Onward Assist’s cloud-based platform introduces certainty into the decision making with these three solutions:
The Path Assist, a full-stack AI platform, covers five biomarkers. “A pathologist looks at specific stained slides of the tissues. As the image is fairly large, it is not possible to look at it in a reasonable time. Machine learning algorithm looks at cell clusters and puts together the information, which helps the pathologist arrive at the right conclusion faster,” Koka states.
The AI-based tools help in slide analysis and the ML algorithm helps in tumor identification, classification, scoring and subtyping. “We also build certain set of models that looks at the same set of samples and extract more clinical insights. The new clinical information with prognostic value helps oncologists predict the accuracy of treatment,” he states.
Rad assist: The ML tools help radiologists detect malignant nodules in mammograms and lung imaging. The tumor assessment helps predict the right treatment and its accurate effect on the patient.
Tumor Board AI: The ML models take cognizance of patient information and predicts how a patient will respond to a particular treatment and modifications in the plan. It also helps segment population into different groups and target high risk groups to improve their results.
Challenges in the process:
Koka says, “Building DeepTech solutions in healthcare takes time. There are image recognition models available but most don’t have anything in common with healthcare. Therefore, a lot of work is fundamental.
An AI model is made to look at a biopsy, teaching it to look at a particular type of cell. A lot of work includes throwing data at a deep learning model and analyzing the result but it also includes making things explainable. The time taken is in bringing the know-how.
Also, making the data available is a process. It is therefore important to collaborate with entities to build curated deidentified data. The platform covers 5 biomarkers, hence the cycle time [initial due diligence of should we work on this, zeroing down on a process, model building, validations] is 18-24 months. It is a time taking process.”
Though the usual suspects—data availability, patient capital and right talent are a challenge, we have come a long way in terms of DeepTech awareness of what is needed to make it successful, believes Koka. The effort therefore needs to be around DeepTech ecosystem and evangelizing it.