This article summarizes three concepts that are/will soon be familiar to anyone interested in clinical trials. These include 1. the idea of an artificial patient; 2. the use of synthetic data; and 3. real-time analysis.
The common thread between the three is that they all have potential tools to speed up, make clinical trials cheaper, safer and more efficient. In an ideal world, this would be true. There is always a catch. We’ll get to that later.
Artificial patient
What does it mean? As of today, there is no final, widely accepted definition of what an artificial/virtual/synthetic patient is. This study provides a detailed explanation of some, but not all, of the various definitions.
Let me summarize what I mean in this analysis.
Artificial patients are data that represent the desired human characteristics in the best way possible. They are based on large quantities of real patient data without any backtraceable real-patient data.
It is why? Artificial patients may be the solution to many problems in modern medicine. Privacy is one of these. A.I. is becoming more sophisticated with the use of machine learning and deep-learning models. A.I. requires huge amounts of data in order to be able to learn. However, giving away a lot of patient data is against their privacy rights. We have seen numerous examples of how dangerous it is to give random companies access to a lot sensitive health data. However, it is possible to take a real-life data set of people and create a synthetic dataset that looks exactly like the original (*more later), without including any personal information.
What’s it good for? Artificial Patients can be used for many purposes, including medical education and clinical trials. This time, we will only focus on the latter. Virtual patients could one day be the most important tool for medical education.
- Estimating the efficiency and potential side effects for promising drug molecules, or optimizing their use.
- To predict the success of future medical devices and treatment methods.
- They can also be used as a substitute for the placebo control group in clinical trials
Artificial patients could one day replace humans and animals, as many believe.
Although artificial patients are promising for drug and medical device development, it is still a very long way before models reach the required complexity and represent the human population.
Artificial patients, which were used as a placebo control group, have also arrived. AppliedVR conducted a VR trial for chronic back pain patients recently. Instead of asking patients to sign up to the trial to not receive treatment (being the control), they decided instead to use a existing patient database, which was provided by Komodo Health, a healthcare data company.
Real-world data can be used as a patient group in research trials, also known as a synthetic arm. This is because companies don’t need to enroll as many people in clinical trial enrollments and can guarantee that all applicants will receive treatment.
Equity in clinical research can be improved by using synthetic control groups. Web Sun, president of Komodo Health and co-founder says that this allows them to look at different subpopulations or underrepresented patient populations in order to determine if they have different outcomes.
Another example is the virtual trial, or in-silico study, which was done to determine the effectiveness of flow diverters for brain aneurysms. Based on data from flow diverter trials, researchers created 82 virtual patients. The experiment predicted an 82.9% success rate using diverters. This is very close to the success rates of three real-world flow diversion trials which had 86.8%, 74.88%, and 76.8% respectively.
Synthetic data
You might feel like your head is spinning because of the idea of an artificial patient. Synthetic data may be even more wild.
What’s it? It is simply synthetic data. To create data sets that are similar to the real world.
Why? We don’t have enough data. To feed any algorithm that requires large amounts of data. To learn new patterns and develop prediction capabilities. HTML2_ A variety of industries and segments use synthetic data, including medicine, self-driving cars, security, robotics and fraud protection, as well as in insurance models and military.
Artificial intelligence is a key component of many medical fields. It can recognise patterns, support diagnoses, and set up treatment paths to optimize healthcare logistics. Intelligent algorithms are able to analyze large amounts of data in a way that no human can and derive clear trends.
Privacy concerns limit the number of data available in medicine. It is difficult to work with sensitive patient information. We cannot both keep our privacy intact and still enjoy the benefits of A.I. in our care. In many cases, sensitive information can be leaked even if it is not intended. We did see this in many cases. Although federated learning, which may make it possible to do so without compromising patients’ privacy is a new method that might be available, its scope is limited.
This is where synthetic data comes in. This can replace the data that is missing, making it possible for completely fabricated patient databases that are as useful in training A.I. It can be used to create a fake patient dataset, but it does not expose any patient data.
This could be overcome by using synthetic data. The training could concentrate on these variables and make use of real-world settings. The above example shows how to diagnose melanoma in patients with dark skin tones. This is a new approach to melanoma diagnosis, which has been difficult to achieve using previous algorithms.
However, not all sunshine and beaches are the same
This detailed analysis not only explains why synthetic data is being used but also why it should scare us. In other words, any dataset that we create will have flaws to some degree. It will have biases that we don’t know about. It won’t include important variables that we have either forgotten about or are unaware of. Even in the best cases, it will only be a snapshot of one moment in a given situation.
Let machine learning and deep-learning algorithms work with these imperfect, synthetic datasets to make sure they come up with conclusions that are closer to the truth in real life.
This is because synthetic data is rapidly spreading. Gartner predicts that 60% of AI data will be synthetic by 2024. Not just for medicine.
Synthetic data is much cheaper and easier to obtain than messy real-world data. What happens when decisions that affect large numbers of people or entire societies are made on the basis of synthetic data?
This is particularly worrying given the challenges facing “truth” in today’s world. Alternative truths that are based on data can be used to support decisions that affect societies, such as healthcare funding and insurance models. This could have disastrous consequences.
Clinical trials decentralised and real-time
What’s it? Use of electronic health data/records/devices for clinical trials. Patients are not required to be present at the site.
It is important. Real-time trials provide faster results, and participants can connect with other patients to share their experiences and gain access to the results.
What’s it good for? More involved participants and quicker results.
In recent months, we have seen some great examples.
Royal-Philips has launched a new home ECG system to support decentralised clinical trials. This new technology will allow clinical trial participants to capture ECG data at home without the need for an in-home physician or travel to a clinical location. Trial participants’ data can be sent to the cloud servers in near real time for analysis.
Many digital health companies are developing tools that enable decentralised trials. Swift Medical, a digital wound-care company, launched a new platform for digital imaging that will allow decentralised clinical trials. This technology is designed to allow large-scale image collection, management and analysis to help researchers monitor the effects of medical interventions at different sites or at home.
The deployment of such advanced technologies in clinical trial will require pharmaceutical companies and biotech companies not only to make financial commitments but also to the belief that technology can significantly help clinical trials. This could include making drugs cheaper, speeding up the process, and making it more convenient for patients.
The Medical Futurist published The Future Of Clinical Trials: Artificial Patients and Synthetic Data.
——————————————————————————————————————————————-
This blog is a source of general information and discussion on health and related topics. Information and materials on this blog, on the website, or in any of the connected materials are not intended to replace or used as a substitute for the advice of a medical professional, diagnosis, or treatment. This blog does not represent the application of any nursing, medical or other health professional advice or diagnosis. We are unable to diagnose health conditions, offer second opinions or provide specific treatment recommendations via this blog or on our website.
If you or another person is suffering from a medical issue and you are concerned, consult your doctor or seek out other medical professional treatment as soon as possible. Do not disregard medical advice from a professional or delay seeking it due to information you seen on the blog or website or in any of the linked materials. If you're experiencing an emergency medical situation, dial 911 or seek emergency medical assistance on the closest phone immediately.
——————————————————————————————————————————————
By: Andrea Koncz
Title: The Future Of Clinical Trials: Artificial Patients, Synthetic Data And Real-Time Analysis
Sourced From: medicalfuturist.com/the-future-of-clinical-trials-artificial-patients-synthetic-data-and-real-time-analysis%ef%bf%bc
Published Date: Tue, 31 May 2022 08:00:00 +0000
Leave a Reply