The 15 th virtual meeting for the IC INNOVATOR CLUB was organized on February 11, 2023 on the topic
‘SUPPORTING VALIDATION OF NEW TECHNOLOGIES IN CLINICAL ENVIRONMENT’
Dr. Vasanth Venugopal, Chief Medical Officer, Clinical Project Lead at CARPL.ai and Radiologist Delhi, India was the guest speaker. Sachin Gaur, Director at InnovatioCuris, opened the session by explaining the importance of clinical validation in building confidence among clinicians for accepting the new built technology. Dr. Vasanth, a technology enthusiast, explained his transition from being a clinician to being an entrepreneur where he built CARPL.ai, a company focused on AI in radiology. In the initial segment, he described the primary role of
- AI in Healthcare
- Personalized Medicine
- Enhanced Medical Imaging (Advancements in deep learning and convolutional neural network).
He further emphasised on AI in Healthcare under the subdivision of:
- Improved diagnosis and treatment planning
- Personalized medicine
- Enhanced medical imaging
- Streamlined clinical decision making
- Predictive analysis for population health
- Reduced healthcare costs through automation
The interception of technology has helped a lot during COVID-19 pandemic as the predictions of the number of mortality and the number of patients who are affected during this wave was made using deep learning tools and their overall approach for future progression was nearly correct as the surge of new variants was also interfering with the studies. These applications helped the authorities to curtail the expense as per the vision of the peaks that narrowed the spread of the disease.
Further wearable technology and sensors was also discussed with suitable examples like Smart watches,smart goggles,smart socks,smart toilets and all the day-to-day activities. There were case studies pertaining to managing the chronic conditions like diabetes which is managed by a patch that has been put subcutaneously and used for two weeks helps in regulating the glucose tracking and monitoring. This patch also help in tracking the hypoglycemic levels and typically avoids hypoglycemic episodes that happen during sleeps as most of the patients end up dying in their sleep due to sudden rise in their sugar level.Physicians do remote monitoring of elderly patients as well as ICU patients today through various upgraded technology and there are times when patients can be monitiored by the physician from their home through monitoring sensors. These technological upgrades are also helpful for disabled and handicapped people. Case studies pertaining to virtual reality (VR) and augmented reality(AR) categories the medical professionals are being trained more extensively and more deeply
More than 500 AI applications have received FDA approval, and 400 companies worldwide are already using AI in their practices in the radiology department.
To begin with validation process he informed the main requirements and principles for validation:
- Patient safety
- Building trust among clinicians and patients
- Effective technology
- Limitations of technology
- Compliance with regulatory bodies.
The validation process: Pre-clinical, Cinical, Post-market Validation. He explained five basic rules for getting approval for validation in clinical environments
1. Strong value, Strong evidence from pre-clinical trials, and clear structure of the data. (Preclinical validation- retrospective modelling, computer simulations, laboratory or animal testing). The technology built/ Company that needs validation should be clear with- Whether the technology affects the final outcome of patients, Reduction in health care cost, Revenue Model, R&D Department performances, Whether these solutions will help them with the problem of unstructured data.
A case study from CARPL.ai was discussed on the transitioned from retrospective settings to clinical validation. Old data from hospitals were taken and labeled using Tech Stack by CARPL, ran them on the solution which gave them the result under a huge bucket of classification of diseases, the solution used several statistical performance matrices.
2. Design Study – the overall strategy will be utilized to carry out a trial that defines a succinct and logical plan to tackle and have desired outcome.
3. The need to be flexible and find ways to do trials in limited resources. (Compare automated AI tools with semi automated available technology to help clinicians know the efficiency.)
(Build a strategy to work efficiently)
4. The right set of tools should be provided to clinicians, linking validation with their workflow, with a user-friendly process and transparent with intuitive results (Helping clinicians know what happens behind the result screen).
5. The validation screen should have a design hack that explains failures and informs clinicians about the solution to deal with limitations. (Outcome + framework that captures the input on a similar screen / In a single interface.)
A case study was discussed on how the existing workflow cannot be avoided to get affected during a prospective study.
CARPL.ai did the study in an hospital’s radiology department, one radiologist was asked to take radiographs and after interpretation the data was given to the company. The data of the whole day was tested again on CARPL’s solution, and another radiologist was asked to check their validity and reliability, saving time and being no hindrance in the workflow.
In The last part Challenges that take place during the validation phase were discussed
1. The speed and time required for the validation process should be considered
2. Ethical considerations and the participation of all stakeholders.
Last part was the Q&A round,
First question asked was: The Lesson to tool builders for reaching the clinical validation phase and not limiting themselves to published papers to which DR Vasanth suggested to peers to start with the problem, find out what clinician’s problem and then build a solution.
Second Question, Framework for medical technology to be launched – the framework for – It involves obtaining approval from the Central Drugs Standard Control Organization (CDSCO), which is the national regulatory body for medical devices and drugs.
Third Question, while digital therapeutics have been introduced in several countries, including the United States and Europe, the development and adoption of digital therapeutics in India are still at a nascent stage. To which answer was – there is a need for regulatory frameworks that can govern the use of digital therapeutics in India. There are challenges related to the infrastructure required for the implementation of digital therapeutics.
Fourth Question, why not in India AI solutions has not been adopted yet– In India, the current incremental value provided by digital solutions to radiologists is limited to only 2-3%. This results in poor financial returns for radiologists. Additionally, the unit economics of deploying such solutions vary across different markets, and the necessary infrastructure to support these solutions, such as cloud systems and RAS systems, is often lacking in hospitals.
Fifth Question, was by Harmanjot a higher secondary student working on an IOT model sensors to monitor Mental Health (a prototype has been developed and his next plan is to get clinical trial – He was advised to find out research studies published on similar designs to know evidence about effectiveness and safety of prototype, what kind of research studies has been done and then move further.)
Sixth Question, how to build a good data set to help solutions to test their technology to which he answered that the government can play a vital role in facilitating the effective utilization of data by providing policy clarity on data sharing with the public and promoting the creation of structured data.
Last Question, how to deal with subjectivity when comparing demarcations done by solutions and Humans. The subjectivity in picture related comes in cognition phase of demarcating outlines of any organ. The subjectivity can be avoided by settings hard and soft ground rule. Now, if a lesion is being worked on the first point of test would be biopsy and other tests then proceeding with demarcation for segmentation then we can follow hard ground rule of working with cognitions and using solutions later.
The session was ended by thankyou note by Sachin Gaur where Dr Vasanth was asked to mentor for clinical validation in upcoming conference of Innovatiocuris (Topic – Clinical Validation). We all are thankful to Dr. V.K Singh and team Innovatiocuris for organizing such an informative session and making the people aware about the importance of validation in the field of clinical environment.