Disclosure control of machine learning models from trusted research environments (TRE): New challenges and opportunities.

Introduction: Artificial intelligence (AI) applications in healthcare and medicine have increased in recent years. To enable access to personal data, Trusted Research Environments (TREs) (otherwise known as Safe Havens) provide safe and secure environments in which researchers can access sensitive personal data and develop AI (in particular machine learning (ML)) models. However, currently few TREs support the training of ML models in part due to a gap in the practical decision-making guidance for TREs in handling model disclosure. Specifically, the training of ML models creates a need to disclose new types of outputs from TREs. Although TREs have clear policies for the disclosure of statistical outputs, the extent to which trained models can leak personal training data once released is not well understood. Background: We review, for a general audience, different types of ML models and their applicability within healthcare. We explain the outputs from training a ML model and how trained ML models can be vulnerable to external attacks to discover personal data encoded within the model. Risks: We present the challenges for disclosure control of trained ML models in the context of training and exporting models from TREs. We provide insights and analyse methods that could be introduced within TREs to mitigate the risk of privacy breaches when disclosing trained models. Discussion: Although specific guidelines and policies exist for statistical disclosure controls in TREs, they do not satisfactorily address these new types of output requests; i.e., trained ML models. There is significant potential for new interdisciplinary research opportunities in developing and adapting policies and tools for safely disclosing ML outputs from TREs.

Machine learning models, trusted research environments and UK health data: ensuring a safe and beneficial future for AI development in healthcare.

Digitalisation of health and the use of health data in artificial intelligence, and machine learning (ML), including for applications that will then in turn be used in healthcare are major themes permeating current UK and other countries' healthcare systems and policies. Obtaining rich and representative data is key for robust ML development, and UK health data sets are particularly attractive sources for this. However, ensuring that such research and development is in the public interest, produces public benefit and preserves privacy are key challenges. Trusted research environments (TREs) are positioned as a way of balancing the diverging interests in healthcare data research with privacy and public benefit. Using TRE data to train ML models presents various challenges to the balance previously struck between these societal interests, which have hitherto not been discussed in the literature. These challenges include the possibility of personal data being disclosed in ML models, the dynamic nature of ML models and how public benefit may be (re)conceived in this context. For ML research to be facilitated using UK health data, TREs and others involved in the UK health data policy ecosystem need to be aware of these issues and work to address them in order to continue to ensure a 'safe' health and care data environment that truly serves the public.

When innovation outpaces regulations: The legal challenges for direct-to-patient supply of investigational medicinal products.

AIMS: We profile the lack of specific regulation for direct-to-patient postal supply (DTP) of clinical trial medications (investigational medicinal products, IMPs) calling for increased efficiency of patient-centred multi-country remote clinical trials. METHODS: Questionnaires emailed to 28 European Economic Area (EEA) Medical Product Licensing Authorities (MPLAs) and Swissmedic MPLA were analysed in 2019/2020. The questionnaire asked whether DTP of IMPs was legal, followed by comparative legal analysis profiling relevant national civil and criminal liability provisions in 30 European jurisdictions (including The Netherlands), finally summarising accessible COVID-19-related guidance in searches of 30 official MPLA websites in January 2021. RESULTS: Twenty MPLAs responded. Twelve consented to response publication in 2021. DTP was not widely authorised, though different phrases were used to explain this. Our legal review of national laws in 29 EEA jurisdictions and Switzerland did not identify any specific sanctions for DTP of IMPs; however, we identified potential national civil and criminal liability provisions. Switzerland provides legal clarity where DTP of IMPs is conditionally legal. MPLA webpage searches for COVID-19 guidance noted conditional acceptance by 19 MPLAs. CONCLUSIONS: Specific national legislation authorising DTP of IMPs, defining IMP categories, and conditions permitting the postage and delivery by courier in an EEA-wide clinical trial, would support innovative patient-centred research for multi-country remote clinical trials. Despite it appearing more acceptable to do this between EU Member States, provided each EU MPLA and ethics board authorises it, temporary Covid-19 restrictions in national Good Clinical Practice (GCP) guidance discourages innovative research into the safety and effectiveness of clinical trial medications.

Pharmacokinetics of amphotericin B lipid complex in critically ill patients undergoing continuous venovenous haemodiafiltration.

The objective of this study was to examine the effect of continuous venovenous haemodiafiltration (CVVHDF) on the pharmacokinetics of amphotericin B (AmB) in critically ill patients following administration of amphotericin B lipid complex (ABLC). Plasma and ultrafiltrate (UF) samples were collected from patients administered ABLC and either receiving or not receiving CVVHDF. Pharmacokinetic (PK) analysis was performed on eight profiles from patients receiving CVVHDF and six profiles from patients not receiving CVVHDF. For patients receiving CVVHDF, the following median PK data were calculated: area under the concentration-time curve (AUC) = 13.9 h·µg/mL, volume of distribution at steady state (V(ss)) = 1476L and drug clearance (CL) = 27.4 L/h; for patients not receiving CVVHDF, the corresponding median PK data were 11.5 h µg/mL, 2048 L and 43.7 L/h, respectively. The median half-lives calculated during the dosage interval (t(1/2int)) were 30.9 h and 32.5 h on and off CVVHDF, respectively, and the total range of t(1/2int) values was 15.6-180.4 h. Observed median peak concentrations on Day 1 were 0.563 µg/mL and 0.468 µg/mL in patients on and off CVVHDF, respectively. From AmB present in the UF, clearance via CVVHDF contributed<1% of total plasma clearance. The AmB concentration-time profiles for patients administered ABLC on and off CVVHDF were compared and no statistically significant differences in AUC, CL, t(1/2int) and V(ss) were observed. In conclusion, CVVHDF had no clinically significant effect on the pharmacokinetics of AmB following administration of ABLC.