Accurate Prediction of <i>θ</i> (Lower Critical Solution Temperature) in Polymer Solutions Based on 3D Descriptors and Artificial Neural Networks

The ability to trigger changes to material properties with external stimuli, so‐called “smart” behavior, has enabled novel technologies for a wide range of healthcare applications. Response to small changes in temperature is particularly attractive, where material transformations may be triggered by contact with the human body. Thermoreversible gelators are materials where warming triggers reversible phase change from low viscosity polymer solution to a gel state. These systems can be generated by the exploitation of macromolecules with lower critical solution temperatures included in their architectures. The resultant materials are attractive for topical and mucosal drug delivery, as well as for injectables. In addition, the materials are attractive for tissue engineering and 3D printing. The fundamental science underpinning these systems is described, along with progress in each class of material and their applications. Significant opportunities exist in the fundamental understanding of how polymer chemistry and nanoscience describe the performance of these systems and guide the rational design of novel systems. Furthermore, barriers to translating technologies must be addressed, for example, rigorous toxicological evaluation is rarely conducted. As such, applications remain tied to narrow fields, and advancements will be made where the existing knowledge in these areas may be applied to novel problems of science.

show abstract

“…Models of this type have been described in organic solvents, and success is likely when applying this approach to LCSTs in water. [ 45–47 ]…”

Section: Thermodynamics Of the Lcstmentioning

confidence: 99%

Polymers Exhibiting Lower Critical Solution Temperatures as a Route to Thermoreversible Gelators for Healthcare

Cook

Haddow

Kirton

et al. 2020

Adv Funct Materials

144

113

View full text Add to dashboard Cite

show abstract

“…This leads to a linear 10-factor model which shows approximately the same predictive power as that developed by Afantitis et al (R 2 = 0.8874, R 2 cv = 0.8658, s = 24.57). In a further paper, Xu et al investigated an even larger descriptor set and the use of neural networks for the prediction of LCSTs [155]. The researchers showed that the development of an initial linear model on the basis of the enlarged descriptor space does not lead to significant improvements in predictive ability in comparison to earlier work, but that the use of neural networks can lead to further improvements in the predictive ability of a model (R 2 = 0.9625, s = 13.43 for the training and R 2 = 0.9524 for the test set).…”

Section: Lower Critical Solution Temperature (Lcst)mentioning

confidence: 99%

Polymer Informatics

Adams

2010

Polymer Libraries

View full text Add to dashboard Cite

Polymers are arguably the most important set of materials in common use. The increasing adoption of both combinatorial as well as high-throughput approaches, coupled with an increasing amount of interdisciplinarity, has wrought tremendous change in the field of polymer science. Yet the informatics tools required to support and further enhance these changes are almost completely absent. In the first part of the chapter, a critical analysis of the challenges facing modern polymer informatics is provided. It is argued, that most of the problems facing the field today are rooted in the current scholarly communication process and the way in which chemists and polymer scientists handle and publish data. Furthermore, the chapter reviews existing modes of representing and communicating polymer information and discusses the impact, which the emergence of semantic technologies will have on the way in which scientific and polymer data is published and transmitted. In the second part, a review of the use of informatics tools for the prediction of polymer properties and in silico design of polymers is offered.

show abstract

“…For example, machine learning models trained on quantum-mechanical data have been used to screen dielectric properties of polymers , and predict atomization energies of molecules, and molecular dynamics simulations of molten and crystal silicon have been conducted with on-the-fly machine learning of quantum-mechanical forces . In addition, ANNs have been used to successfully predict the lower critical solution temperature of polymers, the glass transition temperature, − and various other polymer properties. − …”

Section: Introductionmentioning

confidence: 99%

Neural-Network-Biased Genetic Algorithms for Materials Design: Evolutionary Algorithms That Learn

et al. 2017

View full text Add to dashboard Cite

Machine learning has the potential to dramatically accelerate high-throughput approaches to materials design, as demonstrated by successes in biomolecular design and hard materials design. However, in the search for new soft materials exhibiting properties and performance beyond those previously achieved, machine learning approaches are frequently limited by two shortcomings. First, because they are intrinsically interpolative, they are better suited to the optimization of properties within the known range of accessible behavior than to the discovery of new materials with extremal behavior. Second, they require large pre-existing data sets, which are frequently unavailable and prohibitively expensive to produce. Here we describe a new strategy, the neural-network-biased genetic algorithm (NBGA), for combining genetic algorithms, machine learning, and high-throughput computation or experiment to discover materials with extremal properties in the absence of pre-existing data. Within this strategy, predictions from a progressively constructed artificial neural network are employed to bias the evolution of a genetic algorithm, with fitness evaluations performed via direct simulation or experiment. In effect, this strategy gives the evolutionary algorithm the ability to "learn" and draw inferences from its experience to accelerate the evolutionary process. We test this algorithm against several standard optimization problems and polymer design problems and demonstrate that it matches and typically exceeds the efficiency and reproducibility of standard approaches including a direct-evaluation genetic algorithm and a neural-network-evaluated genetic algorithm. The success of this algorithm in a range of test problems indicates that the NBGA provides a robust strategy for employing informatics-accelerated high-throughput methods to accelerate materials design in the absence of pre-existing data.

show abstract

Accurate Prediction of θ (Lower Critical Solution Temperature) in Polymer Solutions Based on 3D Descriptors and Artificial Neural Networks

Cited by 15 publications

References 45 publications

Polymers Exhibiting Lower Critical Solution Temperatures as a Route to Thermoreversible Gelators for Healthcare

Polymers Exhibiting Lower Critical Solution Temperatures as a Route to Thermoreversible Gelators for Healthcare

Polymer Informatics

Neural-Network-Biased Genetic Algorithms for Materials Design: Evolutionary Algorithms That Learn

Contact Info

Product

Resources

About