|
 In a major relief for digital health innovators and medical technology developers, the Central Drugs Standard Control Organisation (CDSCO) has introduced a streamlined framework for managing algorithm modifications in artificial intelligence and machine learning (AI/ML) based medical device software (MDSW).
The regulatory clarity comes as part of the final Guidance Document on Medical Device Software issued under the Medical Devices Rules (MDR), 2017. The apex drug regulator’s latest move directly addresses long-standing industry concerns regarding the regulatory burden associated with frequent software updates and continuous learning loops inherent to artificial intelligence.
Prior to this guidance, software developers faced ambiguity regarding whether incremental updates to AI models required fresh regulatory approvals, leading to potential delays in deploying software improvements. Under the new framework, the CDSCO has established a clear differentiation between major and minor software modifications, allowing minor algorithm adjustments that do not alter the intended use or core clinical functionality to be managed under internal Quality Management System (QMS) change control protocols. This distinction provides manufacturers with a predictable pathway to maintain software performance without filing full re-licensing applications for routine updates.
Central to this regulatory update is the formalization of the Algorithm Change Protocol (ACP), which requires developers to define predefined modification boundaries during the initial license submission. By documenting expected learning parameters and acceptable performance bounds in advance, manufacturers can deploy pre-approved algorithm changes automatically or through routine software updates. The protocol ensures that continuous learning models operate within established safety margins while preserving clinical efficacy.
The guidance further specifies that if an algorithm update expands the intended diagnostic scope, alters risk categorization, or introduces new clinical decision algorithms, it will be treated as a major change requiring formal submission to the licensing authority. Developers must maintain complete traceability of all code changes, retraining datasets, and performance evaluations to substantiate that algorithm adjustments remain compliant with safety requirements.
Industry stakeholders, including med-tech startups and digital diagnostics companies, have welcomed the ACP framework as a progressive step toward facilitating digital health adoption in India. The clear guidance balances rapid technological innovation with rigorous safety oversight, ensuring that patient safety is not compromised while allowing state-of-the-art diagnostic algorithms to reach clinical settings faster.
The regulator has emphasized that the framework aligns India’s regulatory approach for software as a medical device with globally harmonized practices, particularly those adopted by international regulatory bodies. By harmonizing modification rules, the CDSCO aims to reduce regulatory friction for international medical software manufacturers entering the Indian healthcare market.
Drugs Controller General of India (DCGI) Dr. Rajeev Singh Raghuvanshi advised all digital health developers and medical device manufacturers to integrate the ACP guidelines into their software development lifecycles. The regulator stressed that adherence to these structured update protocols will streamline regulatory reviews, enhance product transparency, and foster sustainable growth across India's digital health ecosystem.
|