Comment from ACM Global Labs

all of these comments are high-level with the hope to have a more detailed discussion. We believe that an innovative central lab should have an opportunity to be included in these discussions. All of the Challenges and Potential AI solutions in the RFI need accurate and QC data from clinical trials. Whether this comes from a local laboratory or a central laboratory, this needs to seen first by the principal investigators at the clinical site to ensure that the trial is managed properly. So, the aspect of QC or adjudicated data needs to be taken into account when assessing what the definition of “real-time” actually is. Regulators should be concerned with the data that will be submitted at the end of the trial in order to achieve time and cost savings, vs the data that might be used for proper management of the study. On the aspect of patient recruitment, I would like to talk through a model of integrating inclusion criteria for patients and integrating this with RWD/RWE to be used at time of visit with the most important intersection of the patient journey, the physician’s visit. Ideally, this is a trusted provider that is managing the care of the patient and is best suited to help identify criteria and willingness of the patient to participate in the trial and with the comfort of provider support be better likely to stay enrolled in the trial itself, thus reducing costs. On the aspects of the pilot project and design, here are my comments. 1 a-b the AI LLM that will be chosen will be critical to the type of trials that will be included. I would expect that one would want to start off with the least complicated trials and as human-in-the-loop confirms the LLM’s ability, complexity and stage of trial design can be increased. A1 c all aspects have potential benefit, however the risk level will be different at each level of oversight and needs to account for human verification at critical steps that could affect patient safety. A2 a willingness to participate would be the first criteria, along with ability to scale to larger patients sets and willingness to integrate. The public good over profit should also be a consideration with any public-private partnership. A2 b ensuring that the smaller more nimble and innovative companies are represented will allow for new ideas and avoid group thing of large systems that may have limitations needed to be overcome for advancement of the pilot objectives. A3 a public-private partnerships I view as the most effective as they have the opportunity to create a new operating model and offer value to the patient, private innovators and government entity as a whole. A3 b beyond comments for the RFI, opening up the ability for smaller companies and/or individuals with innovative ideas to have a voice is crucial to developing the most effective output model and not just a solution that aligns with the chosen vendors strengths. A3 c at a minimum patient advocacy groups that focus on safety and privacy should be of paramount importance in this process. A4 a The FDA should provide clear end-state desire of the pilot and also ensure safety and privacy guardrails are adhered to. The ‘how’ of solutions should be incombered the least as to allow for the greatest innovation and allow for a large-scale change vs small incremental improvements. A4 b-c a central data-base that is secured by government oversight along with secure porting, API’s / ETL’s should be developed to allow for a central risk-point from a security perspective. Providing these solutions will allow non-IT innovators to participate in other areas of their expertise without having to invest in IT security. A5 a-b it is difficult to assess without looking at the projected project pilot details and understanding where there is greatest need for discussion and innovation, of which those should have the longest time-period for collaboration. A5 c Unlike biological models, the evaluation of AI/Digital success should be able to be known through user-testing and security vulnerability testing on the developed systems. B1 a-c understanding the current data-set that the FDA has to measure average time and cost to completion should be use to assess the benefits of this pilot program vs. legacy solutions. B2 a-c similar answer to B1(a-c). 3 a current reporting timelines could be used to determine baseline and ability to identify sooner, which may lead to new criteria should this improve patient safety 3(b-c) – One would think the level of inclusion (ability to capture more/all AE’s), level of detail of AE reporting and increased quantitative data of AE’s should be used to assess better-than-current-state. The remaining of the questions should be addressed based on the solutions remit and then aligned with the current oversight criteria this is used in these areas. T.J. Mueller Chief Strategy and Commercial Officer ACM Global Laboratories