Ram Kumar Mishra
Share
  •  
  •  
  •  
  •  

Poor adherence to prescribed drug dosage regimen during clinical trials can significantly undermine the validity of results, increase costs, and make it difficult to assess the true impact of the medication. But, the reality is, a large percentage of patients participating in a clinical trial do not take their drugs according to the defined protocol due to a variety of reasons – ranging from simple forgetfulness to fear of adverse side effects. According to a study, a 40% patient non-adherence in Phase 4 clinical trials leads to an increase of $6 million in costs for enrolling additional patients to achieve statistical validity. Using the same metrics, a 1% improvement in adherence leads to cost savings of $167,050. It’s clear that the clinical trials industry needs effective tools to monitor dosage and reduce the overall cost. Artificial intelligence (AI) based monitoring platforms promise disruptive solutions to dramatically improve patients’ adherence.

A US company AICure created an app using an AI-based platform and tested it on participants undergoing anticoagulation therapy. The app, installed on participant’s phone, requires the patient to take a selfie while ingesting the pill. The captured video is analyzed using AI to verify whether the patient actually took the medication. Apart from helping clinicians understand whether a patient took the medication as prescribed, it also reminds the patient to consume the drug at the right time. The results were highly promising: adherence in the intervention group using the app was 100%.

In another Phase 2 trial of a nicotinic receptor agonist (ABT-126) in participants with schizophrenia, adherence among patients using the AICure platform was 25% higher than those who were directly monitored. In both studies, clinicians were not only able to monitor participants individually, but were also able to identify poor-performing subjects to improve data quality.

AI in clinical trials leads to cheaper, faster, and more successful trials

An AI platform leverages facial recognition in combination with electronic patient reported outcomes (ePRO) to ensure that accurate data is collected for all patient-reported outcomes. It provides study teams with real-time data for review and intervention. By detecting non-adherence early on in the trial, the platform enables immediate follow up, and higher rates of adherence and completion. Higher adherence and drug concentration increases the statistical validity of data, leading to cheaper, faster, and more successful trials.

A paper published by researchers from Carnegie Mellon University and the University of Freiburg says that machine learning lowers uncertainty in trials, thereby reducing the number of rounds required, by using algorithms that select meaningful experiments based on emerging patterns. According to the paper, an algorithm helped develop a model that was accurate 92% of the time in a case in which only 29% of the experiments needed for the study were conducted. The research proved that a series of experiments using machine learning was feasible even when their set of outcomes was unknown. With these methods, it might also be possible to reduce the costs for achieving the results of multi-site projects. A UK-based start-up BenevolentAI had announced plans to test such a technology by mid-2017.

AI-based platforms drive the SIMPLE approach for better outcomes

Figure 1: SIMPLE approach to increase patient adherence

Strategies to improve patient adherence can be represented using the SIMPLE approach as shown in Figure 1. AI provides patients with a time-stamped recording, while AI-based chat-bots clear patient doubts and fears through effective counselling. By reducing bias and imparting knowledge to patients, AI-based platforms drive the SIMPLE approach for superior outcomes.

Personalized medicine using AI is not far off

With nearly $300 billion wasted due to patient non-adherence, speeding up clinical trials and increasing their accuracy can dramatically reduce costs and improve public health. AI-based apps provide a user-friendly, scalable, and easy-to deploy solution for both clinicians and physicians. Increasing patient compliance and the accuracy of the collected data not only improves the of validity clinical outcomes, but also enhances decision-making by determining ineffective drugs early on to accelerate clinical trial results.

With ever-increasing computational capabilities of smartphones and other internet-enabled devices, AI is poised to accelerate the development of personalized medicines. What do you think? Are you planning to leverage AI to improve clinical practices and clinical trials in your organization?


Share
  •  
  •  
  •  
  •