In this segment, we'll give you a brief introduction to clinical trials where I discuss some of the math underlying the design of this last week, but here we are going to look at the actual design and implementation of this type of studies. All filial trust in the United States have to be preregistered on this webpage that the NIH grants called clinicaltrials.gov, and you can search for what trials are ongoing. This is useful for patients who are interested in participating if they have some disease versus no treatment, you may be able to enroll in a clinical trial and potentially get access to some therapy ahead of time. But part of the reason for this preregistration here is this issue of p-hacking. You can imagine that you have a situation where a company is going to do fault clinical trials for it's own track, in the old days, you could do those four trials and only report the results of a successful clinical trial. Just hide the other three. By pre-registering, by being forced to declare what trial you're going to do ahead of time and registered and then edge webpage, somebody can find out exactly how many trials you are ongoing and ask you well what happened to the other three? Why you only reporting this one? This create some more integrity into the system. The formal definition of a clinical trial, it's a ratio study in which one or more human subjects are prospectively assigned to one or more interventions, which may include placebo rather control to evaluate the effects of those interventions on health-related biomedical or behavioral outcomes. This is the core of it, we're going to walk through a little bit of the terms and especially this was prospective and the word randomized in the next slide. What is prospective and what is a randomized? A prospective clinical trial is where the study design, the outcome specification, the analysis plan are fully defined prior to any data collection. The plan is created, it's reviewed, and it's set in stone. This avoids the temptation to adjust the conditions once experimental process underway which may introduce biases in the process. We set up the conditions and just follow the plan and that's it. Randomization and we talked about this last week, the subjects must be randomly assigned to each group to avoid biases in subjects assignment. This is a two pillars of when you have a randomized prospective clinical trial that is the gold standard. How do we test medical devices on patients? We have a chicken and egg problem, we need approval prior to clinical use and we need clinical testing prior to approval. Somehow we have to break this chain and the solution is a clinical trial which involves experimental use of the device under either an institutional review board approval or an investigational device exemption ID from the FDA. If you're trying a device that doesn't have clearance for use in clinical settings, you either go to your institutional review board, every major hospital has one of these things, you submit your study and you specify all these other risk and there well controls a small number of patients, we know what we're doing and the risks, we have specified them, we have all the checks in place, please approve it. If the level of risk rises above a certain level, you need to go to the FDA itself and to gain, to carry an investigational divine exception granted to you to be able to perform the studies with this device. The same applies to drugs, the same applies to all things here. An idea is required for US human studies that have not been approved for the provost task, which is all of them here in the post significant risk to the patient where the institutional isn't willing to take the risk to approve it effectively, and you need to go through a regulatory approval process. There are two ideas, there's what's called the feasibility study, is a small study to establish the safety of the device and the potential for effectiveness. Safety first, this is safe and it's likely to work and this should follow the usual preclinical studies. For example, animals studies, phantom studies, all the things you do before you think about country humans. The next step is a pivotal study. This is a larger study and this really serves as a primary clinical support for the full FDA application. This is used to demonstrate reasonable assurance of safety and effectiveness, and the endpoints and the numbers of subjects are driven by prior analysis, this is the power analysis we talked about last week. The FDA review of this is more involved. Effectively when you apply for a pivotal study, you set the numbers in place, the set-up in place such that if it succeeds, then from here is a straightforward process for the FDA to give you approval for the device to be using clinical care. This is the last step prior to approval. Now, in this pandemic era, you may have followed the drug studies and the vaccines studies and how either review of those has happened, the process you have to use is similar for drug studies. We have a Phase 1 study, it is a safety testing in a small group, we give the drug small number of people make sure there no side effects. This is Phase 1. Phase 2, we have a larger group divided into a control and experimental group, and here we determined safety and at least an initial hint of effectiveness. This is a Phase 2, at the end of Phase 2 if successful, you go to a Phase 3. This a large and multi-site trial and this is the last step prior to requesting approval from the FDA to introduce a drug to the market. Those of you who follow the vaccine trials, you hear Pfizer announced the results of its Phase 3 trial, sometime in November and then FDA at least initial approval followed later that much. The phase 3 is the large step before getting approved. But we can have issues with drugs later, and their post-market trials that can be performed later to evaluate long-term safety and this is the next part that sometimes happens. This is a process for drug studies and it's very similar to devices and remember software is a device. As we think about testing things on people, we have to worry about the ethical side of things and what does it involved, and one of the key documents is Helsinki Declaration of 1964, we're going to read a couple of paragraphs. This sets out some of the principles used when we test new things on people. Medical research, this goes to the declaration is subject to ethical standards that promote and ensure respect for all human subjects and protected health and rights. While the primary purpose of medical this is to generate new knowledge, this goal can never take precedence over the rights and interests of individual research subjects. The declaration here and you hear the echoes of this comes after World War 2, where we had this horrific scientific experiments done and Nazi concentration camps, I had the privilege to visit Dachau Concentration, come outside of Munich and if you visit and hear about what was done in the name of science, you appreciate why we had the need to develop this type of documents and why this type of symbols are there. Again, remember that in every case, the need for new knowledge can never take precedence over the rights and interests of individual research subjects studies, the fundamental consent and this by the way was issued by the World Medical Association. That is the unlock on the top left of this slide. What does this mean in practice? There are two components, the first one is consent. We must ensure that the subjects are provided informed consent, that they know what risk they're entering into and that they approve they understand what they're getting into. It involves clearly explaining to the potential risks and benefits if there are any benefits to the patient, and the informed consent forms are required to be signed by the patients after detailed explanations provided by the study investigators. We don't take advantage of our patients and ask them to undergo some experimental procedure without explaining to them in language they can understand both the actual language, English, Spanish, whatever is appropriate, and the simplicity of the language which have to be able to appreciate what they're getting into, that is a critical component. The second ethical consideration when it comes to subject recruitment is diversity, and we have to consider the ethical considerations in the determination of the study population. This may be because we want to protect certain groups from harm for example, we may not include children in some studies, and to ensure that the study populations are diverse enough to ensure applicability to both sexes, multiple ethnic groups, differently groups as appropriate. This is particularly important, we're validating AIML modules that are very sensitive to the data, and this type of device also needed to be able to identify and quantify differential performance in different groups. For example, you have a software that detect skin cancer from images, if the software has worst performance in people with darker skin, at the very least willing to be able to inform patients and doctors that this is the [inaudible]. Be careful with people from a particular ethnic group, or truth doesn't work as well on them, this is something you need to do extra examinations on. The other reason that ethical component comes here is to avoid targeting people for higher-risk operations, particularly there's a soda history of targeting particular ethnic group for experimentation in the United States and elsewhere over the past few years. This concludes our discussion of the introduction to clinical trials, in the next segment, we'll take a look at a little bit more of the nuts and bolts of what it takes to design a validation plan. Thank you.
