東華大學圖書館 |

Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence./
作者:	Mahri, Mohammed.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2019,
面頁冊數:	144 p.
附註:	Source: Masters Abstracts International, Volume: 82-10.
Contained By:	Masters Abstracts International82-10.
標題:	Artificial intelligence. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28266921
ISBN:	9798708714442

Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence.
Mahri, Mohammed.

Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence. - Ann Arbor : ProQuest Dissertations & Theses, 2019 - 144 p.

Source: Masters Abstracts International, Volume: 82-10.

Thesis (M.Sc.)--McGill University (Canada), 2019.

This item must not be sold to any third party vendors.

IntroductionThere is growing evidence associating patient systemic conditions and medications to the success of osseointegrated medical devices such as dental implants and hip prosthesis. However, bibliographic assessment of these associations cannot be fully achieved with conventional systematic reviews due to the broad scope of the question addressed. Evidence mapping methods are better suited to such a task; however, evidence mapping can be very resource-intensive. Artificial intelligence can be used to reduce the workload associated with systematic reviews (SR) and systematic mappings (SM). However, the available methods are limited in their ability to reduce the workload and their sensitivity and specificity. A limiting factor is the quality of the training datasets used for machine learning.HypothesisSystematic mapping of the effect of medications on bone-implant osseointegration can be successfully achieved using a machine learning (ML) algorithm trained with similar and nonsimilar training datasets.ObjectiveThe objective of this study was to develop a method for systematic mapping of the literature using a machine learning algorithm trained with similar and non-similar training datasets and use this to identify the effect of medications on bone-implant osseointegration.MethodsTo produce high-quality training datasets for machine learning, we conducted precise search strategies to produce similar and non-similar articles using PubMed. The articles were screened manually and classified into include and excluded articles. The inclusion criteria were clinical and animal studies that assessed the effect of systemic medication on bone-implant osseointegration.The dataset of included and excluded articles screened manually were used to train a machine-learning algorithm based on Support Vector Machines (SVM). The algorithm produced was validated against a published systematic review with a search strategy that falls within the scope of ours. Then, the trained algorithm was used to screen articles identified with a highly sensitive search strategy (543927 articles).ResultsOur algorithm was able to screen half-million published articles and reduce the workload by 95% with an accuracy of 95%, a False Positive Rate (TFP) of 95%, a sensitivity of 93%, and a specificity of 95%. The number of articles retrieved and included for the final analysis was 268 articles. In these articles, we identified 31 drug families that have been studied for their effect on osseointegration.ConclusionPartial automation of systematic mappings can be successfully achieved with similar and nonsimilar training datasets classified by MeSH-terms. This method allowed us to perform a systematic mapping on the effect of medications on bone-implant osseointegration, and we identified 31 drugs that affect osseointegration.

ISBN: 9798708714442Subjects--Topical Terms:

516317
Artificial intelligence.
Subjects--Index Terms:

Osseointegration pharmacology

Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence.
LDR:07306nmm a2200361 4500 001 2347050
005 20220711065235.5
008 241004s2019 ||||||||||||||||| ||eng d
020 $a 9798708714442
035 $a (MiAaPQ)AAI28266921
035 $a (MiAaPQ)McGill_6395wc99f
035 $a AAI28266921
040 $a MiAaPQ $c MiAaPQ
100 1 $a Mahri, Mohammed. $3 3686260
245 1 0 $a Osseointegration Pharmacology: A Systematic Mapping Using Artificial Intelligence.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2019
300 $a 144 p.
500 $a Source: Masters Abstracts International, Volume: 82-10.
500 $a Advisor: Marino, Faleh Tamimi.
502 $a Thesis (M.Sc.)--McGill University (Canada), 2019.
506 $a This item must not be sold to any third party vendors.
520 $a IntroductionThere is growing evidence associating patient systemic conditions and medications to the success of osseointegrated medical devices such as dental implants and hip prosthesis. However, bibliographic assessment of these associations cannot be fully achieved with conventional systematic reviews due to the broad scope of the question addressed. Evidence mapping methods are better suited to such a task; however, evidence mapping can be very resource-intensive. Artificial intelligence can be used to reduce the workload associated with systematic reviews (SR) and systematic mappings (SM). However, the available methods are limited in their ability to reduce the workload and their sensitivity and specificity. A limiting factor is the quality of the training datasets used for machine learning.HypothesisSystematic mapping of the effect of medications on bone-implant osseointegration can be successfully achieved using a machine learning (ML) algorithm trained with similar and nonsimilar training datasets.ObjectiveThe objective of this study was to develop a method for systematic mapping of the literature using a machine learning algorithm trained with similar and non-similar training datasets and use this to identify the effect of medications on bone-implant osseointegration.MethodsTo produce high-quality training datasets for machine learning, we conducted precise search strategies to produce similar and non-similar articles using PubMed. The articles were screened manually and classified into include and excluded articles. The inclusion criteria were clinical and animal studies that assessed the effect of systemic medication on bone-implant osseointegration.The dataset of included and excluded articles screened manually were used to train a machine-learning algorithm based on Support Vector Machines (SVM). The algorithm produced was validated against a published systematic review with a search strategy that falls within the scope of ours. Then, the trained algorithm was used to screen articles identified with a highly sensitive search strategy (543927 articles).ResultsOur algorithm was able to screen half-million published articles and reduce the workload by 95% with an accuracy of 95%, a False Positive Rate (TFP) of 95%, a sensitivity of 93%, and a specificity of 95%. The number of articles retrieved and included for the final analysis was 268 articles. In these articles, we identified 31 drug families that have been studied for their effect on osseointegration.ConclusionPartial automation of systematic mappings can be successfully achieved with similar and nonsimilar training datasets classified by MeSH-terms. This method allowed us to perform a systematic mapping on the effect of medications on bone-implant osseointegration, and we identified 31 drugs that affect osseointegration.
520 $a Introduction: Il existe un nombre croissant de publication associant les patients polymediques a un risque plus important de non osteointegration des implants dentaires et des protheses de hanche. Cependant, la litterature contient un nombre tres important de publication sur le sujet, ce qui rend l'analyse systematiques tres complique. Recemment, des methodes de cartographie de la litterature (ou mapping review) ont ete propose pour realiser ce genre de synthese. Cependant, ce type de travail necessite beaucoup de temps et de ressources. Ainsi, l'intelligence artificielle pourrait etre utilisee pour reduire la charge de travail demande lors de la realisation de ce type de cartographies systematiques. Les methodes disponibles sont actuellement limitees en termes de performance, notamment en termes de sensibilite et leur specificite. Ces performances s'expliquent principalement par la qualite et le nombre de donnees utilises pendant la phase d'apprentissage de l'algorithme.Hypothese: Nous pensons qu'il est possible de realiser une cartographie systematique de l'effet des medicaments sur l'osteointegration des implants osseux en utilisant un algorithme d'apprentissage automatique forme avec des donnees de formation similaires et non similaires. Objectif : Lors de ce travail, nous souhaitons developper une methode de cartographie systematique de la litterature a l'aide d'un algorithme d'apprentissage automatique forme a partir d'ensembles de donnees de formation similaires et non similaires, et de l'utiliser pour identifier l'effet des medicaments sur l'osteointegration des implants en os.Materiels et Methodes: Afin de produire des articles similaires et non similaires, un protocole de recherche precis a ete developpe pour extraire des articles a partir de la base de donnees PubMed. Les articles ont d'abord ete tries et classes manuellement pour rechercher les articles similaires et non similaires. Les criteres d'inclusion etaient des etudes cliniques et animales evaluant l'effet d'un medicament systemique sur l'osteointegration des implants osseux. Les articles inclus et exclus ont ete utilises pour former un algorithme d'apprentissage automatique base sur des machines a vecteurs de support. L'algorithme a ete ensuite valide par comparaison avec une revue systematique prealablement publiee. Enfin, l'algorithme a ete utilise pour selectionner les articles identifies par une strategie de recherche extremement sensible.Resultats: L'algorithme a ete capable d'analyser un demi-million d'articles publies et de reduire la charge de travail de 93% avec une precision de 95%, un taux de faux positifs (TFP) de 95%, une sensibilite de 93% et une specificite de 95%, en comparaison avec la revue systematique deja publiee. Le nombre d'articles recuperes et inclus pour l'analyse finale etait de 266 articles. Dans ces articles, nous avons identifie 31 familles de medicaments qui ont ete etudies pour leur effet sur l'osteointegration.Conclusion: Ce travail a permis de creer un algorithme capable d'identifier et de selectionner avec succes un ensemble d'article a partir des termes MeSH , avec une precision tres proche de celle realise par le travail prealablement. Cette methode nous a permis de realiser une cartographie systematique de l'effet 31 medicaments sur l'osteointegration des implants osseux.
590 $a School code: 0781.
650 4 $a Artificial intelligence. $3 516317
650 4 $a Systematic biology. $3 3173492
650 4 $a Pharmacology. $3 634543
653 $a Osseointegration pharmacology
653 $a Systematic mapping
690 $a 0800
690 $a 0423
690 $a 0419
710 2 $a McGill University (Canada). $3 1018122
773 0 $t Masters Abstracts International $g 82-10.
790 $a 0781
791 $a M.Sc.
792 $a 2019
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28266921