This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 8 minutes read

Trends in digital health: Open letter to O'Shaughnessy Review

This article is the fifth in our "Trends in digital health" series.

Lord O’Shaughnessy has been asked to make recommendations to the UK Government to make the UK a more attractive jurisdiction in which to conduct commercial clinical trials.

AI technologies present one of the most exciting opportunities to radically accelerate the development of novel therapies and medical technologies. As such, we have focused on AI technologies in this open letter to Lord O’Shaughnessy.

AI Opportunities

There are three significant challenges in the current process of identifying and developing a new medicine, each of which could be significantly advanced using artificial intelligence (AI) technologies such as machine learning (ML):

  1. Identifying the optimal therapeutic Target in the patient;
  2. Designing a therapeutic Agent (molecule, antibody or cell therapy) to interact with the Target; and
  3. Determining which patients are most likely to respond positively to that Agent.

The following provides a useful example of how AI has already been deployed to assist with the third of these challenges.

In 2022, Exscientia reported that the results of the EXALT1 clinical trial had demonstrated that AI-mediated functional precision oncology can improve patient outcome in a prospective interventional trial. Exscientia’s AI technology was used to analyse biopsies of each patient’s “normal” and cancer cells to seek to identify a medicine (or combination of medicines) that might work for each patient. This involved:

(a) Dividing the cells into numerous individual samples, each of which was exposed to various medicines (and combinations of medicines); and

(b) Monitoring each sample using robotic automation and computer vision, which in turn used ML models trained to identify small changes in the cells in response to the medicines.

The system was designed to model the way in which each patient might respond to various different medicines. Ideally the system would find a medicine that would have the greatest effect on the patient’s cancer cells without causing too much damage to the patient’s healthy cells.

One patient enrolled in the trial had an aggressive form of cancer that had resisted six separate courses of chemotherapy. This traditional “trial and error” approach was not working for the patient. The medicine that the system recommended was not approved for the patient’s type of cancer. The patient was administered the medicine and two years later the patient is in remission.

The differentiating factor in the AI-led approach adopted by Exscientia is the ability to test all the potential treatments at the same time. The resulting acceleration in finding the right treatment is potentially a gamechanger for patients such as the one in the example. The AI led approach replicates what the doctors treating the patient were doing, with the key difference that the doctors’ approach necessarily involved one treatment at a time. The effect of this ‘in parallel’ approach is loosely analogous to what we have seen during and since Covid in relation to the development of vaccines; if you can carry out certain processes simultaneously, rather than sequentially, treatment times (or in the case of vaccines, development times) can radically shorten.

Exscientia does not want to stop at merely the identification of the best existing medicine to treat a patient. It wants to design (and trial) entirely new medicines. Since 2021, two investigational medicines that Exscientia has developed (or co-developed) have started clinical trials. Exscientia is preparing applications to commence clinical trials of two more novel medicines.

The promise of AI-mediated drug discovery and development is real. However, it requires access to patient-level data and biological specimens. This requires navigation of the requirements of GDPR and laws regarding the storage and use of Human Tissue in research.

Dear Lord O’Shaughnessy,

We respectfully suggest that you consider the following changes to the UK landscape for clinical studies.

Specialist review team for novel or challenging applications 

New and emerging technologies, including AI technologies, can significantly accelerate the development of novel therapies and medical technologies. Focusing on AI, such technology can:

  1. Dramatically accelerate drug discovery through the identification of therapeutic Targets and the design of therapeutic Agents to interact with those Targets;
  2. Assist with the identification and enrolment of clinical study participants who are likely to benefit from the therapy being investigated in the clinical study;
  3. Assist with the analysis of clinical study data to identify statistically significant therapeutic benefits;
  4. Be a novel medical technology in itself – for instance, AI has already been put to powerful use in companion diagnostics, analysing complex genetic data to determine the likelihood that a particular patient with a particular tumour will be responsive to a particular type of chemotherapy. 

For the UK to reap the benefits of new and emerging technologies such as AI, authorities which consider applications to conduct clinical studies must have a sound understanding of these technologies; not only of their immense research and therapeutic potential, but also of their genuine limitations.

We suggest the establishment of a specialist team (or teams) which will assist the MHRA and HRA to review applications for research ethics committee opinions and clinical study authorisations. Such a team would be well-placed to comment on the design of clinical studies involving new and emerging technologies, allowing a flexible approach to clinical study design while ensuring that studies are conducted ethically and safely.

Such a team would need to have the freedom to call upon subject matter experts to provide advice. Experts with a deep understanding of the technological issues should enable the team to make rapid and well-informed decisions. We might suggest contributions from independent experts such as the Information Commissioner’s Office and the PHG Foundation.

Such a specialist team could also provide regulatory assistance to clinical study sponsors regarding the drafting of appropriate Patient Information Statements and Informed Consent Forms for clinical study participants. This would reduce regulatory barriers to entry for innovative SMEs (such as university spin-outs) trying to bring new and emerging health technologies to market.

Banking of biological materials gathered in the course of clinical studies

Biological materials gathered in the course of clinical studies present a uniquely valuable resource for researchers. Biological materials are particularly useful to researchers where they are linked to de-identified clinical data regarding the donor patients. Such materials can be a valuable resource for all kinds of research, but also to train new AI and ML algorithms.

Currently, it is very difficult for the sponsor of a clinical study to ensure that it has proper lawful bases under the Human Tissue Act 2004 and GDPR to hold such materials once the study is over and to use such materials in future research projects. We have worked with a significant number of sponsors who have inadvertently failed to comply with these requirements because they are simply difficult to fully comprehend and somewhat contradictory. 

Data Protection

All personal data must be handled in accordance with GDPR, including data which has been pseudonymised using measures such as key-coding. Full anonymisation of personal data which one wants to use for clinical research tends to be impractical because removing all identifying information makes the data significantly less scientifically useful and, in any event, it is extremely difficult completely to anonymise biological specimens which contain genetic information about the donor.

As such, it is necessary to find a lawful basis under GDPR for storage and other processing of samples and data in a biobank, even where the materials have been key-coded to avoid direct identification.

This poses a continuing challenge as consent is not considered an appropriate lawful basis for the processing of personal data gathered in a clinical trial. Rather, most biobanks (such as the UK Biobank) rely on public interest and research purposes or legitimate interests.

This does not closely align with the requirements under the Human Tissue Act 2004.

Human Tissue Act 2004

The Human Tissue Act 2004 presents a compliance challenge. Without seeking to delve too deeply into its complex requirements, most sponsors of clinical studies conducted in the UK[1] which want to continue to hold biological samples obtained from study participants will seek specific consent from the study participants to allow the “relevant material” to be used for future research projects.

It can be difficult to obtain informed consent from study participants. On the one hand, one must provide the participant with enough information to ensure a complete understanding of the intended uses of the material. On the other hand, providing too much information can confuse the participant and jeopardise the validity of the consent. This tension is particularly pronounced where participants are asked to consent to storage of samples for use in future research, because the future research for which the samples could be used is as yet unknown.

The Human Tissue Act does allow for the storage of samples without the need for consent where the research project for which the samples are stored has been approved by an appropriate Research Ethics Committee. However, in practice, researchers find this difficult to rely upon where they wish to store and use tissue samples for potential future research projects once a clinical study is over. Clearly, such future research projects do not yet have ethical approval because they have not yet been devised.

To some extent the difficulties with the Human Tissue Act have been resolved by bodies such as the UK Biobank operating under a generic ethical approval to enable it to collect, store and release tissue. This allows third parties to have access to samples held by such bodies.

Safe Harbour

We suggest the establishment of a clinical research “safe harbour” to permit “banking” and ongoing access to biological samples gathered in the course of clinical studies conducted in the UK. This could take the form of a Code Of Conduct for biological samples (and key-coded data) collected in the course of clinical studies, that would set criteria for a sponsor to establish a repository of such samples (and associated data) for use in future research. This would need to take account of the Human Tissue Act, patient confidentiality and data protection considerations. This guidance would establish a genuine “safe harbour”, ideally supported by statute, to enable the ongoing storage for as yet unknown research studies provided that associated data has been key-coded. We think that a code of conduct under the Codes of Conduct provisions of Article 40 of the UK GDPR could then form an appropriate data protection mechanism to align with and facilitate this suggested approach.

Such an approach would be consistent with the 2020 findings of a public dialogue commissioned by the HRA and the HTA that there is broad public support for greater data sharing with biobanks – provided that donors are given a clear explanation of how their data and tissue samples could be used.

In our view, such steps could make the UK a significantly more attractive jurisdiction in which to conduct clinical studies, which in turn could significantly accelerate the development of novel therapies and medical technologies.


[1] We acknowledge that the laws regarding human tissue are different in Scotland, but we do not want to be distracted by the distinction for these purposes. That disparity in itself presents an unnecessary stumbling block for human tissue compliance.  


digital health, trends in digital health, artificial intelligence, data protection and privacy, health tech, life sciences, life sciences regulatory, robotics, technology