A Comprehensive Look at the Use of AI for Clinical Trials. Disrupting Clinical Development with Artificial Intelligence

The role artificial intelligence (AI) plays in the pharmaceutical industry has grown considerably over recent years. Clinical trials are essential for drug development, and AI technology may provide a revolution just as it has in other sectors. Artificial intelligence’s potential and prospects in clinical studies are enormous. AI can assist a company in several ways, such as patient selection for the clinical trials, cohort composition, recruitment assistance, and monitoring patients throughout a clinical trial. The following article will provide a comprehensive look at how AI is being used to make clinical trials more efficient, safer, and practical, which helps scientists better understand and target diseases and treatments faster than ever before. We also discuss how it is changing how patients participate in research alongside their doctors by providing a more personalized experience with increased data security and a better understanding of a patient’s history.

The Pharma Drug Development Cycle

The drug development cycle is a series of stages a company must go through to develop a drug for a particular disease (also called a molecule). The drug development process is a long and complicated one, and it usually involves several steps.

First, a company must identify a compound they believe will have the desired effect on a disease. This R&D phase usually takes up to six years. You can read more on drug discovery and development and the potential of AI in our recent article. After identifying a potential drug, companies are required to perform several preclinical studies to determine if the product is safe for human use and whether it works (or has efficacy). During this time, just 10 out of the first 10,000 candidates to be tested for new medications make it to clinical trials. The typical duration of a clinical trial is more than five years.

Clinical Trial Phases

Stage 1: Pre-Clinical Development Phase

This stage involves a lot of work and provides a company with the opportunity to test a molecule in vitro (in the laboratory) and in vivo (in animals) to assess the candidate’s safety, potency, and efficacy. This will help assess whether a drug is safe enough to be tested on humans.

This is also the first opportunity to identify potential side effects and define the maximum tolerated dose. The submission to regulatory authorities is the next stage in this process. This is when the pharmaceutical company must notify the appropriate bodies of future human testing.

Stage 2: Clinical Research Phase I

The next phase in clinical trials is a small-scale study of a drug’s effectiveness and safety in humans, called a Phase I trial. This stage usually involves 20 to 100 healthy volunteers (called subjects) who are closely monitored for any harmful effects or side effects caused by the new medication (for oncological treatments, cancer patients with particular diseases are involved).

The primary purpose is to establish dosage protocols before testing more extensive groups of patients with different degrees of severity to determine what dose should be used later when testing a potential treatment for a specific disease or condition. Phase I studies also serve as an early indication of whether a drug has any efficacy at all; it may be found that the hypothesized mechanism does not work in people.

When a molecule enters this phase, it’s a pharmaceutical company’s responsibility to study the drug and ensure its proper safety for human consumption. This is an essential step in a clinical trial because a molecule that doesn’t pass this phase will be thrown out and never see a market release.

Stage 3: Clinical Research Phase II

Once a pharmaceutical company has a candidate that completed Phase I, it can begin the next phase of clinical trials. This stage usually involves a group of 100 - 300 people with a particular disease or condition to determine whether a drug is working and if there are any side effects.

The purpose is to study how well a drug works for its intended use in humans; researchers will closely monitor patients’ symptoms and their response to treatment (i.e., changes in tumors). The dose used during Phase II studies should be similar to what was studied earlier on healthy volunteers but may not match exactly. There still needs more real-world data before deciding upon the best therapeutic strategy. It must show evidence of an effect over a placebo.

Phase II studies are also used as a second chance at finding possible side-effects before the more extensive Phase III trials take place. If a candidate does well enough during its first phase of testing, it can start this next stage where more people with the disease will participate in clinical trials.

Stage 4: Clinical Research Phase III

This stage usually involves a much larger group of people, anywhere from several hundred up to thousands. This phase is focused on testing drugs on patients to assess efficacy, effectiveness, and safety. The goal is to compare the new medicine to existing therapies for the same condition and demonstrate that it is at least as safe and effective as existing treatment options. Usually conducted as a typical double-blind test, neither the participant nor the investigator knows which medication the participant is taking. If demonstrated that the drug is as safe and effective as others already on the market, the FDA will usually approve the medication in the following stage.

Stage 6: Regulatory Approval Phase

At the end of a clinical trial, a company needs to submit a summary report (called a Clinical Trials Protocol), which includes some required information.

The results should show how many patients were enrolled in the study and their demographics, any side effects they experienced, or treatments given during a trial; if it was successful or not, at showing an impact compared with a control group. Such success can be measured through different measurements such as blood pressure reduction etc., but ultimately what matters is whether there’s enough evidence for regulators to approve a drug for marketing purposes.

If approved, pharmaceutical companies will market drugs within their target population on prescription only until it eventually becomes generic.

Stage 7: Post-market Surveillance

The clinical trials a drug passes through before it reaches the market are a rigorous process, but things still can change once a product is on the open market. Once a medicine has been approved, a company will continue the clinical trials at the post-marketing stage. This is known as Phase IV, which investigates any long-term effects on patients who have taken a candidate drug over a more extended period (usually five years or more) and also monitors for side effects or adverse reactions that may not have surfaced during earlier studies on volunteers.

A company needs to pick this up quickly and respond appropriately if something happens that could harm people who take medicine - anything from an allergy reaction to a fatal error with dosage levels. This means there’s also a very stringent regulatory compliance framework in place. It’s worth noting that these regulations are international across all markets where drugs are sold because of concerns around safety for all patients.

The Cost of Clinical Trials

Clinical trials are a long and expensive process, with a typical clinical trial taking around three years to complete. A clinical trial is a significant investment for a pharmaceutical company. According to a 2001 estimate by the Pharmaceutical Research and Manufacturers of America (PhRMA), a single drug can cost up to $500 million throughout its entire research process, from preclinical testing until market release (for a large molecule with a new mechanism). Success rates of getting a new drug to the market are not encouraging 1 in 5000 of drugs that enter preclinical testing end in approval in a process that usually lasts for over ten years.

In the world of computer technology and semiconductors, in particular, we’ve been noticing a trend that the number of transistors on a microchip doubles every two years. However, the cost of computers is halved. Unfortunately, the pharmaceutical industry and drug development, in particular, are facing an opposite trend termed Eroom’s Law. DesEroom’se the fact that pharmaceutical businesses have steadily increased R&D spending for decades, the number of new medicines gaining regulatory approval per billion USD spent has decreased by half every nine years. It’s a long-termIt’sblem that is already costing the existing clinical development model: in the era of the post-blockbuster drug, a lack of go-to-market efficiency of that magnitude (and growing) is unsustainable. The significant failure rate of clinical trials is one of the main issues in the drug development process.

A single trial can cost anywhere from $44 million to $115 million. But here’s the kicker: according to a CenterWatch survey, delays can cost a trial between $600,000 and $8 million per day.

The cost of patient recruitment constitutes a significant part of all clinical research expenses. The US pharma research and development market reaches over $7 billion, and clinical trial recruitment costs are worth $1,89 billion.

Patients are at the heart of these clinical trials, and poor patient recruitment has dire consequences:

• 72% of studies run more than one month behind schedule
• 53% of the time enrolment timeline exceed planned enrollment timelines
• 25% of studies fail because they fail to enroll patients
• 50% of sites enroll one or no patients in studies
• 85% of clinical trials fail to retain enough patients
• 80% of all clinical trials fail to finish on time

For these reasons, it’s crucial for pharma trial sponsors to find the right fit for clinical trials from the start.

How Can Artificial Intelligence Help in Clinical Trials?

The use of artificial intelligence in clinical trials has a lot of potential and a lot of opportunities. It is being used by companies worldwide thanks to its effectiveness, safety benefits (fewer errors), and savings both with money, but it also enables faster advances through research than ever before. Using artificial intelligence solutions allows us to improve our understanding of how diseases work to develop better treatments that are tailored just for you as a patient.

AI can assist a company in achieving its goals across all aspects: from patient selection, monitoring patients during trial, study closeout, and many other places where it provides benefits like new insights or an increase to quality assurance tasks done at faster rates with higher accuracy than ever before.

Protocol Development and Study Design Optimization

Clinical Trial Protocols are a guide for clinical trial conduct to be followed by investigators and monitors. These protocols address essential issues such as patient safety, informed consent of study participants, confidentiality agreements with research subjects, what will be done if a problem occurs during the study (serious adverse events), etc. The protocol should include procedures for any additional or unforeseen circumstances that may arise during a clinical trial.

The cost, efficiency, and success potential of clinical trials are all harmed by poor study design. We may use AI technologies and natural language processing, in particular, to determine and choose optimal primary and secondary endpoints in study design to ensure that the most relevant protocols for regulators, payers, and patients are defined using massive healthcare data sets. More precise study design enables more predictable findings, shorter protocol development cycles, fewer protocol modifications, and greater efficiency throughout the trial. It also leads to improved participant recruitment rates, non-enrolling sites dropped from the study and more occasional protocol changes. These benefits increase the chances of success and help researchers develop more realistic and accurate plans.

Site Identification and Patient Recruitment

AI can also be a great help in site identification and recruitment of patients for a clinical trial. Using several applications, companies will be able to advertise clinical trial opportunities more easily and recruit patients.

AI can assist a company in many ways, bringing savings time and money-wise. For example, AI provides a cost-effective way for organizations to identify viable sites that are readily capable of performing the proposed study activities without requiring significant site modifications or rework based on prerequisites such as geographical location, availability of equipment, and time constraints.

Selecting a patient for a clinical trial is a multistep process that requires several different steps. A company may use a number of applications, including AI platforms, to make this process easier. Using artificial intelligence tools is an excellent way for companies to get their clinical trials started quickly by efficiently identifying patients who fit into specific groups before conducting eligibility screenings with them. This will allow investigators to focus more resources on recruiting participants than screening people out. The first step in the process, identifying possible patients, can be done by reviewing records from previous studies (from other companies) to see if they would qualify and then contacting them about enrolling in your company’s clinical trial.

Structured and unstructured clinical data, including doctor’s notes, electronic health records, pathology reports, medical imaging, ‘omics’ data (e.g., genomics, metabolomics, and proteomics), operating notes, and other vital medical records in free text form or visuals that cannot be easily searched, can all be readily analyzed with custom AI software. Artificial intelligence and natural language processing are used to extract tens of thousands of new clinical data points – symptoms, diagnoses, treatments, genomics, lifestyle data, and more – from fragmented medical documents. The software transforms fragmented medical records into cohesive patient graphs that contain all of the information needed to match complex clinical trial criteria.

Precision analysis of patient populations based on AI and clustering technologies can deconstruct traditional symptom-based categories. Patients with various disorders can be studied across other integrated medical data to parse current heterogeneous syndromes into homogeneous clusters. Machine learning algorithms can analyze the multidimensional data to identify otherwise invisible connections and suggest patient groups for prospective replication and stratified clinical trials. These carefully targeted patient groups are fundamental in the latter clinical research phases to properly evaluate the efficacy of the treatment under research. AI technology applied to massive amounts of patient data can extract meaningful patterns of information to help improve the design of clinical trials. AI makes it possible to use advanced predictive analytics to analyze patients’ genetic material, allowing them to apply sophisticated predictive analytics and help researchers identify the appropriate patient group for a trial.

Finding Clinical Trial

Artificial intelligence can also assist a company in engaging and recruiting new patients. The use of machine learning and natural language processing can help discover publicly accessible web content, including digital trial announcements, trial databases, and social media posts, to assist in matching patients with available clinical trials. For example, AI platforms allow companies to market themselves and their clinical trial opportunities more efficiently through digital channels like social media to engage with patients who are most likely to enroll or respond well to a specific therapy. This will help a company increase the number of people registered for a clinical trial faster than ever before because it allows them to reach out directly to these individuals rather than just trying to find them on traditional advertising methods. We all know how expensive some forms of marketing can be, so using artificial intelligence is an excellent way for pharmaceutical drug development companies, both small and large alike, to save money while still publishing information about what types of therapies they offer.

Patient Engagement

Patient engagement is going to be another area where AI could start seeing increased usage in clinical trials. There are many different ways artificial intelligence can help with patient engagement; from improving the overall experience for patients who sign up and participate in clinical studies, a clinical trial patient is given a health care provider who will be responsible for treating them throughout the course of their participation in a study. This could mean that they’re seeing a doctor one-on-one to discuss their condition, participating in group sessions led by a clinician or other medical professional, and even attending classes where patients can learn about what kinds of treatments are available for specific conditions.

Patient engagement is critical because you don’t want people dropping out before your study has ended, which means it’s necessary to make sure every single person who signs up gets all the support they need during each step of treatment until the end date arrives. Patient retention rates have become increasingly low over time which makes it difficult for companies to produce reliable results when conducting clinical trials, and this problem isn’t limited to just pharmaceutical drug development companies either; even smaller CROs have had retention rates below 20% which means you need almost double the number of people enrolled than expected just so your data isn’t skewed because there aren’t enough people left at the end of a study.

Study Monitoring and Real-time Trial Insights

Artificial intelligence can help a company monitor a clinical trial by collecting and analyzing data in real-time. These AI solutions can collect a lot of information from many sources, which would then enable doctors or researchers to understand how patients react during a drug trial.

Clinical research offers a wealth of operational data, but numerous applications and functional data silos prevent pharma executives from obtaining a complete picture of their clinical trials portfolio. Because of that, hours are lost each day attempting to optimize trial operations and enhance cost and resource efficacy by collecting and analyzing various data sets. By combining operational data from clinical trials with AI and advanced predictive capabilities on an analytics platform, pharmaceutical firms can better determine whether a data aberration is a genuine risk, allowing for more efficient and successful visits.

In addition, it provides insights into the patient’s progress, which has been another potential breakthrough for companies using this technology: these insights have helped them reduce side effects and increase their understanding about what works best for each individual patient when it comes to treatment options.

Data collected during recruitment and ongoing clinical research adds additional value in the future. It has the potential to create personalized trial participation experiences to drive patient adherence and retention during a trial. Insights from the trial help develop patient profiles for future tests to accelerate recruitment and reduce startup costs.

Clinical Data Review and Study Closeout

AI tools are an excellent solution for automated document processing and data extraction. As clinical trials produce massive amounts of documents in different formats, there is always a need to process them intelligently, fast, and accurately. Automated data collection and machine learning platforms can empower clinical teams to perform the most tedious tasks in a fraction of time. This kind of AI-based solution can find any number or combination of identifiers, such as patient ID and date of birth, and enable full-text search through trial documents. They can use NLP to search data and obtain the status of their actions using a self-documenting platform, which provides them with a real-time overview of the clinical study state.

Clinical trials collect data from electronic data capture systems and local laboratories. The resulting information is then kept in a data warehouse. The data is converted into a format that the data managers are comfortable with. Once the data is cleaned, it’s transformed to standard data models and submitted to the FDA for approval.

Machine learning techniques can interpret medical data, analyze it using statistical analysis, and perform a preliminary cleaning of the data. Data reviewers may prioritize their quality assurance efforts based on the clean/dirty probability. Machine learning models can be used to detect patterns in data and notify data managers if there are any anomalies or missing information. ML models may provide greater insight into clinical data and allow people to assess the trial’s safety, adherence, and effectiveness.

Virtual Trials

With the Covid pandemic, many businesses have moved online and remote. The pharmaceutical industry and clinical trials were no different. We could see an increase in the implementation of virtual clinical trials. In virtual trials, patients can be enrolled in real-time and in their usual setting (versus just controlled clinical settings) and remotely monitored. Rather than relying on a single site for their treatments, patients can use the latest remote monitoring technology (telemedicine) to access diverse specialists across the country or the world for more specialized care and take part in virtual trials. This boosts doctor accessibility and reduces costly time commitments that individuals face when dealing with non-electronic medical information systems. Thanks to evolving tech, virtual trials are based on telemedicine solutions and use dedicated apps for patient engagement and tracking or e-signature platforms to lessen the financial and time burdens that patients incur.

Virtual trials may reduce patient dropout and speed up study timelines, which are two of the most pressing challenges in life sciences research and development. In fact, the transition to virtual clinical trials implies that any suitable participant who wishes to participate in medical research and meets the criteria can do so.

Pharmacovigilance

Pharmacovigilance, or PV in short, is a regulatory requirement that pharmaceutical companies must comply with. This system safeguards the public from any potential harm a company may cause by not following its compliance policies and procedures, manufacturing standards or good laboratory practices (GLPs), clinical trial design, protocol execution, and data management requirements.

Pharmacovigilance has three main objectives:

• Understand a product’s safety and monitor adverse events;
• Monitor the effectiveness of a product to ensure continued benefit for patients;
• Assess risk information related to a medicinal product as well as developing prevention strategies if necessary. It does this through pharmacovigilant surveillance systems such as spontaneous reporting, ongoing monitoring during the development process, post-marketing research studies, among others.

AI’s most common use case in pharmacovigilance is natural language processing (NLP) for an automatic or supervised reporting system. A clinical trial or a drug is not always a controlled environment. That is why the reports from healthcare professionals need to be processed as soon as they are received, which can typically happen through a web portal using a simple interface.

NLP allows us to easily search for patterns in these reported cases without needing a lot of manual effort on behalf of staff members who may receive hundreds if not thousands of emails per day so we can monitor patient safety during a clinical research project with ease and efficiency: it will help you look at data faster than ever before and give you more insight into the results than what was available only a few years ago– this technology speeds up your ability to make decisions by providing advanced analytics capabilities such as summarization or data extraction for adverse events and PV terms from electronic health records or literature.

To learn more about PV and AI, head over to Augmenting Drug Safety and Pharmacovigilance Services with Artificial Intelligence (AI) article. To understand the technicalities of NLP better, read our Definitive Guide to Natural Language Processing (NLP).

Robotic Process Automation

Health regulation authorities and the healthcare industry manage vast amounts of administrative work that robotic process automation (RPA) and machine learning could help to reduce.

RPA is computer software that can execute a repetitive task or process automatically. It automates the execution of business tasks such as inputting data into various fields, reading tables and other functions from the screen to an Excel spreadsheet, parsing XML feeds (pulling out specific information), performing calculations using formulas found on a sheet - just some examples of what RPA platform can do.

In clinical trials, robotic process automation can be used to reduce the number of manual tasks a person has to perform, and this can include:

• Recording a patient’s response, or a doctor’s assessment, during a consultation;
• Filling in online forms;
• Taking screenshots for digital signatures;
• Extracting data from electronic medical records;
• Sync and process data from multiple data sources;
• Extraction of information from scanned documents.

A significant advantage for companies when it comes to automation is how much cost savings they will be able to generate by implementing this solution; there are substantial reductions in labor costs because there’s no need for intervention during routine operations. Another vital aspect of RPA systems is their ability to reduce errors: these automated procedures have produced fewer mistakes than manual processes.

Benefits of AI in Clinical Trials

The use of artificial intelligence technologies in clinical development is not only a fantastic opportunity, but it is also now a critical business need, as suggested throughout the article. By lowering trial duration and expenses while improving patient experience, AI technology provides an opportunity for enhancing the clinical trials process.

The use of AI in drug development offers the following benefits to both pharmaceutical companies as well as patients themselves:

• Better accuracy;
• Faster progress and reduced costs of trials;
• Improvements to the enrollment process and optimized cohort composition;
• Reduced side effects;
• Increased patient retention and lower dropout rates;
• Improved patient engagement and adherence;
• Improve patient outcomes and satisfaction scores thanks to patient-centric approach;
• Maximized chances for a successful outcome;
• Actionable insights and enhanced analytics;
• Egalitarian clinical research accessible for all;
• More effective trial planning and reduced time to market;
• Increased compliance.

The Future of Clinical Trials - Not Only AI

However, remember that AI is not a cure-all; it can’t accomplish scientific discoveries or run trials independently. This technology isn’t meant to take the place of scientists, and it couldn’t do so. It’s simply there to help drug development experts by providing insight into both questions and answers that the technology enables us to inquire about. On the one hand, if we are truly serious about changing how we do drug discovery and development, artificial intelligence and machine learning will have to take a leading role. AI and machine learning will play an increasingly more prominent part in not just clinical trials R&D but all aspects of health and wellness as more people use these technologies to help them live better lives.

References

Artificial Intelligence for Clinical Trial Design - Harrer et al., 2019

Intelligent clinical trials - Delloite Insights Report, 2020

Artificial intelligence in managing clinical trial design and conduct: Man and machine still on the learning curve? - Bhatt et al., 2021