REGULATORY UPDATE | ARTIFICIAL INTELLIGENCE (AI) SOFTWARE REGISTRATION DI SINGAPURA

In December 2019, the Health Sciences Authority (HSA) of Singapore’s Ministry of Health released guidelines regarding software registration in Singapore, including a major section dedicated to AI software. The HSA stresses that these measures act as guidelines and are not official regulatory controls.
The guidelines categorize software into four categories: software embedded in medical devices, standalone software, standalone mobile applications and web-based software. Registration applications are submitted according to the the ASEAN Common Submission Dossier Template (CSDT) format and may be prepared from the International Medical Device Regulators Forum’s (IMDRF) Non-In Vitro Diagnostic Medical Device Market Authorization Table of Contents (nIVD MA ToC).

TOTAL PRODUCT LIFE CYCLE (TPLC) AND CONTINUOUS LEARNING AI

HSA recommends that medical device software manufacturers adopt Total Product Life Cycle (TPLC) approach, overseeing requirement management, risk assessment, software verification/validation, change management and traceability through the duration of the software’s life cycle. While reviewing the entire process will be critical to validating the effectiveness of the claims, the HSA will also pay special attention to post market surveillance and ongoing model monitoring activities, especially when it comes to “continuous learning” AI-MDs. “continuous learning” AI-MD.

ARTIFICIAL INTELLIGENCE (AI) adalah suatu teknologi yang memiliki kecerdasan layaknya manusia yang di modelkan didalam mesin dan diprogram seperti halnya manusia dalam memecahkan masalah kognitif seperti pembelajaran, pemecahan masalah, dan pengenalan pola. AI-MEDICAL DEVICE (AI-MD) merupakan kecerdasan buatan yang dimaksudkan untuk membantu dalam penyelidikan, deteksi, diagnosi, pemantauan, pengobatan atau pengelolaan kondisi medis, penyakit, anatomi atau proses fisiologis.

QUALITY

Manufacturers of medical devices and medical software must have a Quality Management System to ensure quality. All medical devices must be designed and manufactured to be safe and perform well throughout the product life cycle.

In the development of AI-MD, there must be a person responsible for the implementer and developer. There are various laws and guidelines that must be complied with such as: Personal Data Protection Act, Human Biomedical Sciences Act and Private Hospitals and Medical Clinics Act
edical device development can refer to the HAS GN-16 guide: Guidance on Essential Principles for Safety and Performance of Medical Devices. AI-Medical Device must be designed to ensure accuracy, reliability, precision, safety and performance while meeting their intended use

There are special considerations for AI-MD regulation such as continuous learning capability, human intervention, model training etc. Once AI-MD is deployed, active monitoring, assessment and tuning is required. To ensure traceability and also implement mechanisms to monitor and assess AI-MD performance, developers and distributors must work closely with implementers and users to avoid having to implement a concept.

MEDICAL SOFTWARE SECURITY

Medical software requires security just like cybersecurity, but there are considerations in a cybersecurity strategy. Serangan cybersecurity strategy can fatally disrupt medical facilities and the functionality of medical equipment, make hospital networks unavailable and delay patient care. Only with competent cybersecurity can the functionality and security of medical devices be effectively protected.

t requires joint efforts from different stakeholders such as: government agencies, manufacturers, healthcare institutions, users of medical devices. To achieve this cybersecurity strategy as monitoring, assessment, mitigation and risk communication requires active participation of all stakeholders in the ecosystem. Manufacturers should consider all potential cybersecurity risks and consider design inputs that can reasonably secure the device and prevent, detect, respond to and, if possible, remediate foreseeable cyber threats.

In addition to running medical software, there is also a Field Safety Corrective Action (FSCA). This action can be taken with awareness of the risks associated with the use of medical devices through post-market monitoring and supervision. The product owner will communicate the risk to the victim and inform what steps will be taken to reduce the risk. To continue corrections without undue delay on the availability of software upgrades or bug fixes.

Bugs introduced in software during the design and development period. There are various causes of software errors, such as failure to check the impact of changes during the software upgrade, incompatibility with installed third-party programs, lack of software validation before the first release, incorrect configuration to perform the corresponding upgrade. operating system.
Problems commonly found in medical software software include inaccurate test results, possible clinical diagnostic errors, calibration errors that lead to incorrect patient positioning, errors in software algorithm calculations that lead to incorrect radiation therapy dose calculations, error alarms such as a software error that causes a false alarm message. to send.

ACTIVITIES CONTROL

For activities to be performed on any party, all manufacturers, importers and wholesalers of medical software software must be licensed for medical devices. The licensee must implement a quality management system (QMS) as a condition, there are several aspects such as ensuring that the software is developed and produced under a good and effective quality management system, ensuring traceability of the medical device software, providing assurance that appropriate procedures are in place for post-market monitoring and response, and ensuring proper maintenance and handling of device-related data and information.