Crowdsourcing in pharma: a strategic framework

Bentzien, Jörg; Bharadwaj, Ragu; Thompson, David C.

doi:10.1016/j.drudis.2015.01.011

Cited by 30 publications

(17 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The kaggle data sets (https://www.kaggle.com/c/MerckActivity) surprisingly did not provide the same set of descriptors for the training and test sets for each data set. Consequently, we selected the set of descriptors common to training and test sets for each data set, so that the test set values could be predicted from the models.…”

Section: Methodsmentioning

confidence: 99%

“…As few DNN QSAR studies have been reported so far, and comparisons of deep and shallow machine learning methods should ideally be done with a relatively large number of ‘real world’ data sets containing large numbers of compounds with high molecular diversity. To this end, the Merck Molecular Activity kaggle competition data set was well suited to such a comparison . The competition allowed a wide range of QSAR modelling approaches, including DNN, to be tested with a consistent, fixed set of molecular descriptors.…”

Section: Comparing Dnn and State‐of‐the‐art Bnn Using Test Casesmentioning

confidence: 99%

See 1 more Smart Citation

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR

Winkler

2016

Molecular Informatics

View full text Add to dashboard Cite

Neural networks have generated valuable Quantitative Structure-Activity/Property Relationships (QSAR/QSPR) models for a wide variety of small molecules and materials properties. They have grown in sophistication and many of their initial problems have been overcome by modern mathematical techniques. QSAR studies have almost always used so-called "shallow" neural networks in which there is a single hidden layer between the input and output layers. Recently, a new and potentially paradigm-shifting type of neural network based on Deep Learning has appeared. Deep learning methods have generated impressive improvements in image and voice recognition, and are now being applied to QSAR and QSAR modelling. This paper describes the differences in approach between deep and shallow neural networks, compares their abilities to predict the properties of test sets for 15 large drug data sets (the kaggle set), discusses the results in terms of the Universal Approximation theorem for neural networks, and describes how DNN may ameliorate or remove troublesome "activity cliffs" in QSAR data sets.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Comparing Dnn and State‐of‐the‐art Bnn Using Test Casesmentioning

confidence: 99%

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR

Winkler

2016

Molecular Informatics

View full text Add to dashboard Cite

show abstract

“…Here, we review a recently published contribution wherein a strategic framework was posited regarding the utilization by an organization of crowdsourcing methodologies [11] . The framework itself provides a unifying view on how organizations could meaningfully engage a group of ‘Others’, and the hope remains that the framework serves as an organizing principle for leaders to coordinate, strategize, and operationalize ongoing ‘networked activities’.…”

Section: Introductionmentioning

confidence: 99%

Crowdsourcing and open innovation in drug discovery: recent contributions and future directions

Thompson¹,

Bentzien

2020

Drug Discovery Today

Self Cite

View full text Add to dashboard Cite

show abstract

“…Many pharmaceutical companies have become adept at defining problems or questions that need to be solved, and they often reach out to the wider scientific community for solutions. A pharmaceutical company typically either establishes its own Internet portal to solicit solutions from potential solvers, as is the case with GlaxoSmithKline (GSK) or Bayer, or works with specialist brokers who have platforms, such as InnoCentive, NineSigma, or Kaggle [ 32 ]. Indeed, the crowdsourcing platform, Grants4Targets, has triggered or generated relevant input into ten drug discovery projects at Bayer [ 33 ].…”

Section: Harnessing the Scientific Crowd Requires A Reliable Framewormentioning

confidence: 99%

Establishing a reliable framework for harnessing the creative power of the scientific crowd

Carter

Donner²,

Lee

et al. 2017

PLoS Biol

View full text Add to dashboard Cite

Discovering new medicines is difficult and increasingly expensive. The pharmaceutical industry has responded to this challenge by embracing open innovation to access external ideas. Historically, partnerships were usually bilateral, and the drug discovery process was shrouded in secrecy. This model is rapidly changing. With the advent of the Internet, drug discovery has become more decentralised, bottom-up, and scalable than ever before. The term open innovation is now accepted as just one of many terms that capture different but overlapping levels of openness in the drug discovery process. Many pharmaceutical companies recognise the advantages of revealing some proprietary information in the form of results, chemical tools, or unsolved problems in return for valuable insights and ideas. For example, such selective revealing can take the form of openly shared chemical tools to explore new biological mechanisms or by publicly admitting what is not known in the form of an open call. The essential ingredient for addressing these problems is access to the wider scientific crowd. The business of crowdsourcing, a form of outsourcing in which individuals or organisations solicit contributions from Internet users to obtain ideas or desired services, has grown significantly to fill this need and takes many forms today. Here, we posit that open-innovation approaches are more successful when they establish a reliable framework for converting creative ideas of the scientific crowd into practice with actionable plans.

show abstract

Crowdsourcing in pharma: a strategic framework

Cited by 30 publications

References 55 publications

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR

Performance of Deep and Shallow Neural Networks, the Universal Approximation Theorem, Activity Cliffs, and QSAR

Crowdsourcing and open innovation in drug discovery: recent contributions and future directions

Establishing a reliable framework for harnessing the creative power of the scientific crowd

Contact Info

Product

Resources

About