Can AI radically increase efficiency in medical technology?
Formulating requirements, structuring information, clarifying ambiguities: A great deal of work is involved in the development of medical devices before the first technical solution is implemented. Requirements gathering, in particular, demands time and care. Product ideas, application scenarios, and technical parameters must be transformed into clear, understandable, and verifiable specifications.
AI can accelerate this process. At the same time, the source information often contains confidential knowledge. This raises a crucial question: How can AI be used productively in medical technology without disclosing sensitive data?
Requirements as an efficiency lever
A good requirement clearly describes what a product or function should do. It creates a common foundation for development and testing. However, formulating it is often time-consuming: information is scattered across meeting notes, existing documents, or the minds of individual stakeholders. It must be compiled and presented in a consistent format.
This is where AI can provide support. It can formulate requirements specifications from provided information and help structure them. This gives the development team a basis for technical work more quickly.
The main benefit lies in reducing repetitive writing and preparation work. More time is then available for questions that require experience and product understanding: Is the requirement correct? Is it complete? And can it be clearly verified later?
Confidential development data needs protection.
Companies using AI for requirements generation in medical technology may be working with information that should not leave the company. This includes, for example, technical product details, customer-specific information, confidential project information, or even personal data.
Therefore, for AI to be used effectively, it must be clear which information may be processed and which must be protected. Equally important is where data is transferred, who can access it, and under what conditions it is stored or further used.
Data sovereignty means consciously controlling how one's own information is handled. This check must be part of the workflow.
AI support with data filtering
At MEDtech Ingenieur, we use AI support to write requirements. Before the information is passed to the AI, a preliminary filter removes sensitive data.
This ensures that confidential information is protected even before AI processing. The processed information then serves as the basis for creating requirements specifications.
We also have such a data filter at Medioora, our requirements tool.
In simplified terms, the process is as follows:
Provide information → filter out sensitive data → create requirements design with AI → review and approve from a technical perspective.
The challenge lies in removing sensitive content while simultaneously preserving sufficient subject-matter context for usable results. Therefore, a carefully designed process also includes testing the effectiveness of the filters and a structured approach to handling unclear input.
A data filter is an important component of data protection. To assess overall data handling, the systems used, access rights, and the AI provider's terms and conditions must also be considered.
Technical responsibility remains with the development team.
A linguistically convincing design is not yet a reliable requirement. AI can misinterpret information, overlook connections, or add content that is not evident from the source data.
Technical review therefore remains essential. Those responsible must assess, in particular, whether a requirement is unambiguous, factually correct, and verifiable. Its consistency with other requirements and its significance for the safety of the device must also be considered.
Changes and approvals should remain traceable throughout the intended development process. This ensures that an AI-generated design becomes a verified working basis.
Consider the actual time savings
The extent to which AI accelerates requirements generation in medical technology becomes clear when looking at the entire process. A quickly created design is of little use if extensive revisions are subsequently required.
For a meaningful evaluation, teams should therefore compare their previous processing time with the AI-supported approach, including data preparation, verification, and rework. It is also important to consider whether the results achieve the required quality.
A clearly defined use case provides a good starting point. Suitable examples can be used to examine what information is required, how reliably sensitive data is protected, and where actual effort can be saved.
With this approach, AI can specifically relieve the burden on medical technology development.
Do you want to accelerate your requirements gathering and protect sensitive development data? Talk to us about the application possibilities of Medioora in your development process.