MLAPI - A framework for developing machine learning-guided drug particle syntheses in automated continuous flow platforms

Aug 5, 2026

Upload by Aria

Recently, machine learning (ML) models have been increasingly used within process analytical technology (PAT) for pharmaceutical manufacturing, but data-hungry requirements limit broader deployment. This work presents a computational, data-efficient framework for guiding drug particle synthesis in an automated continuous-flow precipitation platform. It combines classification to find feasible, fouling-free operating regions, a multi-output Gaussian process model linking key process parameters to drug particle size, and active learning to select informative new experiments. The framework is demonstrated using ibuprofen microparticle synthesis, enabling GP-guided tuning for targeted particle bioavailability and processability.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages14

Reading Time35 min

AccessFree to download

Abstract

Introduction

Bioavailability challenges of solid APIs

Nanoparticle/microparticle formulation and control hurdles

Continuous flow antisolvent precipitation and engineering obstacles

Process Analytical Technology (PAT) for automation and chemometrics

Chemical Engineering Science 302 (2025) 120780 Contents lists available at ScienceDirect Chemical Engineering Science journal [homepage:](homepage: www.elsevier.com/locate/ces)[ www.elsevier.com/locate/ces](homepage: www.elsevier.com/locate/ces) | MLAPI: A framework for developing machine learning-guided drug particle syntheses in automated continuous ﬂow platforms
Arun Pankajakshan a, Sayan Pal a, Nicholas Snead a, Juan Almeida b, Maximilian O. Besenhard a, Shorooq Abukhamees c, Duncan Q.M. Craig d, Asterios Gavriilidis a, Luca Mazzei a,
Federico Galvanin a,∗
a Department of Chemical Engineering, University College London, Torrington Place, London, WC1E 7JE, United Kingdom b Perceptive Engineering, Applied Materials, Vanguard House, Cheshire, WA4 4AB, United Kingdom
c Department of Pharmaceutics and Pharmaceutical Technology, Faculty of Pharmaceutical Sciences, The Hashemite University, Zarqa, 13115, Jordan d Faculty of Science, University of Bath, Claverton Down, Bath, BA2 7AY, United Kingdom | | | --- | --- | | A R T I C L E I N F O
Keywords:
Machine learning
Drug Active Pharmaceutical Ingredient (API) Continuous ﬂow
Automation | A B S T R A C T | | | Recently, machine learning (ML) models are increasingly being used in process analytical technology (PAT) frameworks for pharmaceutical manufacturing. Yet, the applications of ML-integrated PAT frameworks are limited by big data requirements. This work introduces a computational framework to develop data-eﬃcient ML models to guide drug particle synthesis in an automated continuous ﬂow precipitation platform. The framework incorporates classiﬁcation algorithms to identify feasible (fouling-free) operating regions of the precipitation platform, a multiple-output Gaussian process (GP) regression model to relate key process parameters to the drug particle size, and active learning to optimally generate new data for training and validation of the GP model. The usefulness of the proposed framework is demonstrated on the synthesis of ibuprofen microparticles in an automated ﬂow precipitation platform. We envision that properly trained GP models developed using the proposed framework can be employed to ﬁne tune the drug particle size, targeting desired particle bioavailability and processability. | 1. Introduction Poor bioavailability of solid formulations of pharmaceutical drugs caused by low aqueous solubility and dissolution rate of Active Pharmaceutical Ingredients (APIs) is a persistent problem within the pharmaceutical industry. Among several methods to address this problem, nanoparticle or microparticle API formulations are an attractive solution (Mohammadi et al., 2023). Preparing drugs as nano or microsuspensions markedly increases their speciﬁc surface area, speeding up drug dissolution and improving their bioavailability. However, making drugs as sub-micron particles is challenging. It requires signiﬁcant ﬁne-tuning to control particle size, stop particles from clumping together, and prevent the formation of unwanted structures. These challenges have impeded widespread adoption of this technology within the pharmaceutical sector (Jia et al., 2022). Continuous ﬂow antisolvent precipitation (Sinha et al., 2013) is an energy-eﬃcient and scalable approach for the continuous synthesis of drug particles at the nano or micro scale (Jia et al., 2022). From a process standpoint, the main engineering obstacle in producing drug particles continuously is to manage suspensions in ﬂow eﬀectively. Conversely, when considering product quality, the primary engineering challenge in manufacturing drug particles in the sub-micron range is achieving precise control over the size distribution of the drug particles (Benyahia et al., 2012; Lakerveld et al., 2015). This control is vital for ensuring eﬃcient drug delivery. Moreover, the size distribution plays a critical role in various subsequent pharmaceutical procedures, such as ﬁltration, washing, drying,

FAQs

What problem does the framework address in ML-guided pharmaceutical particle synthesis?

It targets the limitation that ML-integrated PAT approaches often require large data sets, which restricts practical adoption.

How does the framework model and guide drug particle synthesis?

It uses classification to identify fouling-free operating regions, a multi-output Gaussian process regression model to relate process parameters to particle size, and active learning to generate new training and validation data efficiently.

What demonstration and application does the paper present?

The framework is demonstrated on ibuprofen microparticle synthesis in an automated flow precipitation platform, with the goal of tuning particle size toward desired bioavailability and processability.

Aug 5, 2026

Upload by Aria