The search for existing non-animal alternative methods for use in experiments is currently challenging because of the lack of both comprehensive structured databases and balanced keyword-based search strategies to mine unstructured textual databases. In this paper we describe 3Ranker, which is a fast, keyword-independent algorithm for finding non-animal alternative methods for use in biomedical research. The 3Ranker algorithm was created by using a machine learning approach, consisting of a Random Forest model built on a dataset of 35 million abstracts and constructed with weak supervision, followed by iterative model improvement with expert curated data. We found a satisfactory trade-off between sensitivity and specificity, with Area Under the Curve (AUC) values ranging from 0.85–0.95. Trials showed that the AI-based classifier was able to identify articles that describe potential alternatives to animal use, among the thousands of articles returned by generic PubMed queries on dermatitis and Parkinson’s disease. Application of the classification models on time series data showed the earlier implementation and acceptance of Three Rs principles in the area of cosmetics and skin research, as compared to the area of neurodegenerative disease research. The 3Ranker algorithm is freely available at www.open3r.org; the future goal is to expand this framework to cover multiple research domains and to enable its broad use by researchers, policymakers, funders and ethical review boards, in order to promote the replacement of animal use in research wherever possible.
van LuijkJLeenaarsMHooijmansC, et al.Towards evidence-based translational research: The pros and cons of conducting systematic reviews of animal studies. ALTEX2013; 30: 256–257.
2.
RatajeskiMAMillerRS. Adapting to changes in publishing when searching for alternatives and reporting on animal research: A librarian’s perspective. Altern Lab Anim2022; 50: 57–61.
3.
de VriesRBHooijmansCRTillemaA, et al.A search filter for increasing the retrieval of animal studies in Embase. Lab Anim2011; 45: 268–270.
4.
YauwSTKWeverKEHoesseiniA, et al.Systematic review of experimental studies on intestinal anastomosis. Br J Surg2015; 102: 726–734.
5.
HabetsPCvan IJzendoornDGVinkersCH, et al.Development and validation of a machine-learning algorithm to predict the relevance of scientific articles within the field of teratology. Reprod Toxicol2022; 113: 150–154.
6.
WangYSohnSLiuS, et al.A clinical text classification paradigm using weak supervision and deep representation. BMC Med Inform Decis Mak2019; 19: 1.
7.
FischerISteigerH-J. Toward automatic evaluation of medical abstracts: The current value of sentiment analysis and machine learning for classification of the importance of PubMed abstracts of randomized trials for stroke. J Stroke Cerebrovasc Dis2020; 29: 105042.
8.
MarshallIJNoel-StorrAKuiperJ, et al.Machine learning for identifying randomized controlled trials: An evaluation and practitioner’s guide. Res Synth Methods2018; 9: 602–614.
9.
Bannach-BrownAPrzybyłaPThomasJ, et al.Machine learning algorithms for systematic review: Reducing workload in a preclinical review of animal studies and reducing human screening error. Syst Rev2019; 8: 23.
10.
Ritskes-HoitingaMAlkemaW. The use of artificial intelligence for the fast and effective identification of Three Rs-based literature. Altern Lab Anim2021; 49: 133–136.
11.
SpeitG. How to assess the mutagenic potential of cosmetic products without animal tests?Mutat Res2009; 678: 108–112.
12.
JohanssonHLindstedtM. Prediction of skin sensitizers using alternative methods to animal experimentation. Basic Clin Pharmacol Toxicol2014; 115: 110–117.
13.
FentemJH. The 19th FRAME Annual Lecture, November 2022: Safer chemicals and sustainable innovation will be achieved by regulatory use of modern safety science, not by more animal testing. Altern Lab Anim2023; 51: 90–101.
14.
CassottaMGeertsHHarbomL, et al.The future of Parkinson’s disease research: A new paradigm of human-specific investigation is necessary… and possible. ALTEX2022; 39: 694–709.
15.
FeinererIHornikKMeyerD. Text mining infrastructure in R. J Stat Softw2008; 25: 1–54.
16.
WickhamHAverickMBryanJ, et al.Welcome to the tidyverse. J Open Source Softw2019; 4: 1686.
17.
LiawAWienerM. Classification and regression by RandomForest. R News2002; 2: 18–22.
18.
RobinXTurckNHainardA, et al.pROC: An open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics2011; 12: 77.
19.
MoudioSRodinFAlbargothyNJ, et al.Exposure of α-synuclein aggregates to organotypic slice cultures recapitulates key molecular features of Parkinson’s disease. Front Neurol2022; 13: 826102.
20.
VecchioLMSullivanPDunnAR, et al.Enhanced tyrosine hydroxylase activity induces oxidative stress, causes accumulation of autotoxic catecholamine metabolites, and augments amphetamine effects in vivo. J Neurochem2021; 158: 960–979.
21.
ZhouQZhangM-MLiuM, et al.LncRNA XIST sponges miR-199a-3p to modulate the Sp1/LRRK2 signal pathway to accelerate Parkinson’s disease progression. Aging2021; 13: 4115–4137.
22.
PuentesLNLengyel-ZhandZReillySW, et al.Evaluation of a low-toxicity PARP inhibitor as a neuroprotective agent for Parkinson’s disease. Mol Neurobiol2021; 58: 3641–3652.
23.
GuptaDPParkSHLeeY-S, et al.Daphne genkwa flower extract promotes the neuroprotective effects of microglia. Phytomedicine2023; 108: 154486.
24.
KarimiAElmiMShiriZ, et al.Therapeutic potential of pluripotent stem cell-derived dopaminergic progenitors in Parkinson’s disease: A systematic review protocol. Syst Rev2021; 10: 188.
25.
European Union. Regulation (EC) No 1223/2009 of the European Parliament and of the Council of 30 November 2009 on cosmetic products. Off J Eur Union2009; L342: 59–209.
26.
van der MierdenSTsaiounKBleichA, et al.Software tools for literature screening in systematic reviews in biomedical research. ALTEX2019; 36: 508–517.
27.
GilmourNReynoldsJPrzybylakK, et al.Next generation risk assessment for skin allergy: Decision making using new approach methodologies. Regul Toxicol Pharmacol2022; 131: 105159.
28.
ShenZSchutteDYiY, et al.Classifying the lifestyle status for Alzheimer’s disease from clinical notes using deep learning with weak supervision. BMC Med Inform Decis Mak2022; 22: 88.
29.
RabbyGBerkaP. Multi-class classification of COVID-19 documents using machine learning algorithms. J Intell Inf Syst2023; 60: 571–591.
30.
Del FiolGMichelsonMIorioA, et al.A deep learning method to automatically identify reports of scientifically rigorous clinical research from the biomedical literature: Comparative analytic study. J Med Internet Res2018; 20: e10281.
31.
MartinSFEsserPR. Avoiding contact allergens: From basic research to the in vitro identification of contact allergens. Allergol Select2017; 1: 77–84.
32.
OtsuboYNishijoTMizumachiH, et al.Adjustment of a no expected sensitization induction level derived from Bayesian network integrated testing strategy for skin sensitization risk assessment. J Toxicol Sci2020; 45: 57–67.
33.
KimberI. The activity of methacrylate esters in skin sensitisation test methods II. A review of complementary and additional analyses. Regul Toxicol Pharmacol2021; 119: 104821.
34.
RussoGCrispinoECorsiniE, et al.Computational modelling and simulation for immunotoxicity prediction induced by skin sensitisers. Comput Struct Biotechnol J2022; 20: 6172–6181.
35.
NatshehHVettoratoETouitouE. Ethosomes for dermal administration of natural active molecules. Curr Pharm Des2019; 25: 2338–2348.
36.
AlbertKNiskanenJKälväläS, et al.Utilising induced pluripotent stem cells in neurodegenerative disease research: Focus on glia. Internat J Mol Sci2021; 22: 4334.
37.
VenkataramanLFairSRMcElroyCA, et al.Modeling neurodegenerative diseases with cerebral organoids and other three-dimensional culture systems: Focus on Alzheimer’s disease. Stem Cell Rev Rep2022; 18: 696–717.
38.
PonomarenkoMSharypovaEDrachkovaI, et al.Unannotated single nucleotide polymorphisms in the TATA box of erythropoiesis genes show in vitro positive involvements in cognitive and mental disorders. BMC Med Genet2020; 21 (Suppl. 1): 165.
39.
TsaiR-TTsaiC-WLiuS-P, et al.Maackiain ameliorates 6-hydroxydopamine and SNCA pathologies by modulating the PINK1/parkin pathway in models of Parkinson’s disease in Caenorhabditis elegans and the SH-SY5Y cell line. Int J Mol Sci2020; 21: 4455.
40.
GądarowskaDKalkaJDaniel-WójcikA, et al.Alternative methods for skin-sensitization assessment. Toxics2022; 10: 740.
41.
Th’eluACatoireSKerdine-RömerS. Immune-competent in vitro co-culture models as an approach for skin sensitisation assessment. Toxicol In Vitro2020; 62: 104691.
42.
TaGHWengC-FLeongMK. In silico prediction of skin sensitization: Quo vadis?Front Pharmacol2021; 12: 655771.
