Leveraging Generative AI in Enhancing Product Owner Responsibilities in the Post-Market Phase of Medical Device Software
Bibliographic record
Abstract
Patients can die from using a medical device if it is not closely monitored after its release into the market. Post-market surveillance (PMS) was introduced by regulators as a requirement for medical device manufacturers to continuously monitor the performance and safety of their devices while they are in the market. However, medical device manufacturers face several challenges in the PMS phase of their devices. But it remained unclear what these challenges are and little research has explored the challenges that device manufacturers face in post-market surveillance. Therefore, the aim of this thesis is to identify the activities carried out by device manufacturers in the post-market surveillance phase and understand the challenges they encounter while monitoring the devices in the market. This thesis also aims to map post-market surveillance activities to scrum product owner responsibilities and suggest ways of using generative AI to simplify the post-market surveillance process. Semi-structured interviews were conducted with four industry professionals who currently have devices in post-market surveillance phase. The study was focused on the European Union with all participants from the European Union. The data collected from these interviews were thematically analysed using Claude 3.5 Sonnet with structured prompts to reveal themes. The study revealed some key challenges in post-market surveillance, including data management and feedback analysis. Currently, there are complexities in managing large amounts of data such as customer feedback, generated during post-market surveillance. These challenges hinder the efficiency of a feedback-driven decision-making process which is crucial for continuous improvement of medical devices. Another crucial challenge was with limited resources, particularly for smaller medical device manufacturers. The inability of manufacturers to sometimes conduct post-market clinical follow-up studies based on customer feedback could hinder the product’s improvement, market expansion, and affect the long-term success of the device. It was also found that generative AI can be potentially used to automate initial feedback processing which could significantly improve the efficiency of categorising and prioritizing feedback. This thesis confirms that generative AI has the potential to improve post-market surveillance of medical devices and offers insights to medical device manufacturers, product owners, project management office roles, regulatory bodies, and AI developers on the application of generative AI in post-market surveillance. Some main limitations of this study include its small sample size of four, its limited scope to the EU, and it only presents a theoretical model.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.028 | 0.072 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.003 | 0.008 |
| Scholarly communication | 0.007 | 0.010 |
| Open science | 0.003 | 0.010 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".