AI Technology

Description

Key outcomes

Ref.

Ligand Competitive Saturation (SILCS) and Machine-learning based random-

Forest (RF)

Screening 700,000 chemicals for the functional group having antimicrobial activity.

Identified CMY-10 inhibitors having potentiality to work against MD.

[62]

Multichannel deep neural network (DeepBLI)

To identify potential metallo—β-lactamaseAIM-1 inhibitors and adapt rottlerin to four different classes of β-lactamase targets, demonstrating its potential as a broad-spectrum inhibitor

Produces results with Area Under the Receiver Operating Characteristics (AUROC) of 0.9240 and AUPRC of 0.9715, indicating the ability to discover novel β-lactamase-inhibitor interactions.

[63]

Machine learning algorithms

A large dataset with more than 62,000 compounds and their β-lactamase potency values AmpC was used in QSAR (quantitative structure-activity relationship) mode after being obtained from ChEMBL.

Microorganisms containing beta-lactamases were tested for bioactivity against plant-based flavonoids and terpenoids.

[64]

DeepBL

The whole proteome was screened using the UniProt database reviewed bacterial protein sequences.

 The discovery of β-lactamase in silico

[65]

Deep Learning-Assisted Photochromic Sensor

For use with convolutional neural networks (CNNs), a dataset of 2520 unduplicated fluorescence intensity images were collected.

The technique permitted quick measurement with a concentration range of 1 to 100 mg/L and distinguished six -Lactams with a prediction accuracy of 97.98%

[66]

Bayes decision rule combined with robust statistics and a Siamese neural network

Detection of targeted Optical Distribution Network (ODN)

Detected T-ODN having NDM-1 enzyme like activity responsible for β -lactam antibiotics resistance.

[67]

Combined Support-Vector-Machine-Based Virtual Screening and Docking Method

For the virtual screening of IMP-1 metallo-β-lactamase inhibitors, a support vector machine (SVM) and the docking method separate compounds into positives and negatives. For in vitro tests, eight of the twenty-five chosen compounds were bought.

Four compounds have demonstrated inhibitory potency against IMP-1 Metallo-β-lactamase inhibitors.

[68]

Plasmonic nanosensors and Machine Learning

Detect nanoparticle surface plasmon resonance (SPR) spectra

Rapid detection of (E. faecium, S. aureus, K. pneumoniae, A. baumannii, P. aeruginosa, and Enterobacter spp.) ESKAPE pathogens.

[69]

ML-random forest model

Using ML to look for and find potential candidates with the properties of β-lactamase inhibition

A search for structurally related compounds was conducted after one molecule that showed much promise, and the results revealed that all 28 of the other returning compounds had antibacterial activity.

[62]

Algorithm-based Artificial Neural Network

A feature vector is created by deriving several metrics from the basic structure. Experimentally determined beta-β-lactamase data are gathered and converted into feature vectors.

Using jackknife testing, cross-validation, and independent testing, the predictor's overall accuracy is 99.76%, 96.07%, 94.20%, and 91.65%, respectively.

[70]