Artificial Intelligence (AI) and Machine Learning (ML) are rapidly changing the landscape of drug development and regulatory submissions. In recent years, these technologies have gained significant attention and have been increasingly adopted by pharmaceutical companies and regulatory agencies alike.
Regulatory submissions are a critical aspect of drug development, and they involve submitting a vast amount of data to regulatory agencies for approval of a new drug or medical device. AI and ML can be used to streamline this process and improve the quality of regulatory submissions. Here are some ways AI and ML are being used in regulatory submissions for drug development:
- Predictive Modeling: AI and ML algorithms can be used to build predictive models that can forecast potential risks and outcomes of clinical trials. These models can help optimize the trial design, identify potential safety issues, and improve the accuracy of endpoint predictions.
- Data Mining: AI and ML can be used to analyze large data sets and extract meaningful insights. This can help identify potential safety issues, support decision-making, and optimize study design.
- Natural Language Processing: AI and ML can be used to process unstructured data, such as clinical trial reports, medical literature, and regulatory guidelines. This can help automate data extraction and standardization, which can improve the efficiency of regulatory submissions.
- Image Analysis: AI and ML can be used to analyze medical images and identify patterns or anomalies that may indicate disease or adverse events. This can help improve the accuracy of diagnoses, predict treatment outcomes, and support regulatory decision-making.
- Pharmacovigilance: AI and ML can be used to analyze real-world data and identify potential safety issues associated with a drug or medical device. This can help identify adverse events earlier, improve signal detection, and support risk management.
Overall, AI and ML have the potential to transform the regulatory submissions process and improve the efficiency, accuracy, and quality of drug development. However, challenges remain to be addressed, such as ensuring the reliability and transparency of AI and ML algorithms and addressing concerns around data privacy and security.
ML Methods and Pharmacometrics:
Structural model selection in pharmacometrics followed by random and covariate effects described the linear process. Hence, in the optimization field, a global search method called Genetic Algorithm (GA) can be proposed as a better alternative and create a user-defined “search space” of all the predictive candidate models representing all hypotheses to be tested. Implementation of GA using NONMEM for parameter estimation is becoming a very robust and cost-effective way for population PKPD model development.
Through the supervised ML, observation of features and associated responses can be extrapolated. ML also helps to make predictions for new observations and infer relationships between responses and features.
The machine learning model was able to accurately predict the PD response of individuals who were treated with dosing regimens that were different from the once-daily regimen used in model construction
Use of AI in Quantitative Structure-Activity Relationship (QSAR)
Nowadays, quantitative structure-activity relationship (QSAR) models are widely used to predict toxicities based on chemical structural parameters. since many drugs with unknown mechanisms of action are available, the application of artificial intelligence (AI)-which uses sophisticated algorithms- is increasingly used to predict toxicities. Recently, the QSAR model was applied to determine complex relations between chemical structures and toxicities.
Pharmacometric Model construction using artificial neural network (ANN):
Sometimes, model development is challenging due to the complexity of the physiological mechanism that governs time-dependent changes of drug concentrations in an individual patient and their biological effects.CDER researchers have recently investigated how methods that use artificial neural networks can be applied to modeling problems.ANN is widely used to simulate the time course of a PD response that is not directly related to the drug concentration. Still, rather that develops latently, according to complex biological intermediate steps. To test the capability of ML to model pharmacodynamic in the scenario, the researchers first constructed a mechanistic PK/PD model for a testing drug that had delayed biological response to change in drug concentration
Recurrent neural network (RNN) had shown a great impact on predicting a different kind of sequential data e.g. time sequence of plasma concentration simulated by mechanistic model along with patient baseline PD values. The machine learning model was able to accurately predict the PD response of individuals who were treated with dosing regiments that were different from the once-daily regimen used in model construction
With the enhanced computing power and stronger algorithms developed in the last decade, not only multiscale models but also predictive algorithms based on artificial intelligence are promoted and a new discipline combining these two, computational pharmaceutics has emerged. Using machine learning (ML), large volumes of data can be analyzed systematically to find correlations or quantify the agreement of correlations. ML can also be used in carrying characteristics across the scales, i.e., in the process of information homogenization.
Reference:
https://www.fda.gov/drugs/regulatory-science-research-and-education/new-approach-pharmacometrics-recurrent-neural-networks-modeling-drug-exposure-and-drug-response
https://link.springer.com/article/10.1007/s11095-022-03298-8
https://pubmed.ncbi.nlm.nih.gov/32238631/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6657004/
https://ascpt.onlinelibrary.wiley.com/doi/full/10.1002/cpt.2668
Leave a Reply