Security, privacy and responsibility

We design according to the European principles of data protection, starting from privacy by design: protecting information is part of the architecture, not something added later.

• We aim for minimization: collecting and exposing only the data the purpose requires.
• We design the platform so that the data identifying a patient stays separate from clinical data, with support for pseudonymization.
• For this website, the privacy policy describes in detail which data we process and why.

We take the European regulation on artificial intelligence into account from the design of the predictive modules onwards, knowing that in the medical field the requirements are stricter.

• We set up our work so that, for every model, the purpose it is meant for and the data it was built on are documented.
• We plan to follow how models behave over time, so as to notice if they become less reliable.
• Modules intended to influence clinical decisions are designed with the regulatory pathway that applies to them in view.

To us security is a property of the architecture: who has access, to what, and what trace is left are questions we answer while we design the system.

• We design role-based access, so that everyone sees what their work requires.
• We plan for the traceability of operations, so that what happened, and when, can be reconstructed.
• We plan for the encryption of data, both in transit and at rest.

We design interfaces that are readable and usable by people with different needs and devices, taking into account the international guidelines on web accessibility.

• We take care of contrast and text size, so that reading takes no effort.
• We design so that navigation also works from the keyboard and with assistive technologies.
• We respect the reader's preferences, such as reduced motion.

For the platform, our orientation is an infrastructure located in the European Union, or directly on the premises of the healthcare organization when it asks for that.

• We design the same software to run in the cloud or on the organization's own systems, without tying it to a single provider.
• The infrastructure is defined with each organization, according to the context of use and its rules.
• This website is a separate case: providers and places of processing are described in the privacy policy.

We think of analysis and research functions as a path separate from that of clinical decisions, with rules of its own.

• We design so that researchers work on data stripped of what identifies the patient.
• Extracting a dataset is designed to go through a request and an approval.
• We adopt open standards, so that data from different centers can be compared.

The predictive modules assist the clinician in weighing risks and scenarios: the decision, with the responsibility it carries, stays with the professional.

• Every indicator and every suggestion is designed to be verified and interpreted by the healthcare professional.
• The clinician can always depart from a suggestion: it is their assessment that counts.
• No black boxes: those who use the system must be able to understand what happened and why.