Artificial Intelligence Guided Research and Development

Abstract

Recommendations for new experiments are generated via a pipeline that includes a predictive model and a preference procedure. In one example, a definition of a development task includes experiment parameters that may be varied, the outcomes of interest and the desired goals or specifications. Existing experimental data is used by machine learning algorithms to train a predictive model. The software system generates candidate experiments and uses the trained predictive model to predict the outcomes of the candidate experiments based on their parameters. A merit function (referred to as a preference function) is calculated for the candidate experiments. The preference function is a function of the experiment parameters and/or the predicted outcomes. It may also be a function of features that are derived from these quantities. The candidate experiments are ranked based on the preference function.

What is claimed is: 
     
         1 . A method implemented on a computer system comprising a processor, the processor executing instructions to effect a method for recommending candidate experiments for a development task, the method comprising:
 receiving, via a user interface, user-provided information about a development task, the user-provided information including specifications for the development task and variation of experimental parameters for candidate experiments for the development task;   automatically generating a preference function that is based on the user-provided information about specifications for the development task;   generating a number of candidate experiments based on the user-provided information about variation of the experimental parameters;   calculating preference scores for candidate experiments according to the preference function; and   recommending from among the candidate experiments based at least in part on their preference scores.   
     
     
         2 . The method of  claim 1  further comprising:
 training a predictive model based on known outcomes of previously conducted experiments characterized by known experimental parameters; and 
 applying the predictive model to predict outcomes of the candidate experiments based on their experimental parameters, wherein the preference scores are calculated based on the predicted outcomes. 
 
     
     
         3 . The method of  claim 1  wherein the user-provided information about specifications for the development task comprises priorities for different specifications. 
     
     
         4 . The method of  claim 3  wherein the user interface comprises dropdown menus for selecting priorities for different specifications. 
     
     
         5 . The method of  claim 1  wherein user-provided information about specifications for the development task comprises thresholds and goals for at least some of the specifications. 
     
     
         6 . The method of  claim 1  wherein the preference function is a combination of different components based on different specifications included in the user-provided information. 
     
     
         7 . The method of  claim 1  wherein the user-provided information about specifications for the development task comprises priorities for different specifications; and the preference function is a combination of different components for different specifications where the components are weighted based on priorities of the specifications. 
     
     
         8 . The method of  claim 1  wherein the user-provided information about variation of the experimental parameters comprises variation in composition of ingredients and variation of process parameters for processing the ingredients. 
     
     
         9 . The method of  claim 8  wherein the user interface comprises separate sections for variation in composition of ingredients, and variation of process parameters. 
     
     
         10 . The method of  claim 1  wherein the user-provided information comprises constraints on the experimental parameters. 
     
     
         11 . The method of  claim 10  wherein the user interface comprises a separate section for constraints on the experimental parameters. 
     
     
         12 . The method of  claim 10  wherein constraints on experimental parameters comprise constraints on ratios of different experimental parameters. 
     
     
         13 . The method of  claim 10  wherein constraints on experimental parameters comprise requiring different experimental parameters to be used together and/or prohibiting different experimental parameters to be used together. 
     
     
         14 . The method of  claim 1  wherein the user-provided information further includes information about previously conducted experiments, and generating the candidate experiments is based on the user-provided information about previously conducted experiments. 
     
     
         15 . The method of  claim 14  wherein the information about previously conducted experiments includes composition of ingredients and process parameters for the previously conducted experiments, and the user interface comprises separate sections for the composition of ingredients and the process parameters. 
     
     
         16 . The method of  claim 14  wherein the information about previously conducted experiments includes at least one of: material lots for ingredients, metadata for the material lot, and user notes on experiments. 
     
     
         17 . The method of  claim 1  further comprising:
 providing an interface that allows the user to query by experimental parameter and/or outcome of experiment. 
 
     
     
         18 . The method of  claim 1  wherein the development task is one of: optimizing rubber properties, optimizing paint, optimizing food, optimizing drug delivery, optimizing drug stability, optimizing drug production, optimizing production of a material, optimizing process parameters for a manufacturing line, optimizing production of a part, optimizing a product for consumer preferences, or optimizing variations of a product. 
     
     
         19 . A non-transitory computer-readable storage medium storing executable computer program instructions to effect a method for recommending candidate experiments for a development task, the instructions executable by a processor and causing the processor to perform a method comprising:
 receiving, via a user interface, user-provided information about a development task, the user-provided information including specifications for the development task and variation of experimental parameters for candidate experiments for the development task;   automatically generating a preference function that is based on the user-provided information about specifications for the development task;   generating a number of candidate experiments based on the user-provided information about variation of the experimental parameters;   calculating preference scores for candidate experiments according to the preference function; and   recommending from among the candidate experiments based at least in part on their preference scores.   
     
     
         20 . A system for recommending candidate experiments for a development task, the system comprising:
 a first user interface for receiving user-provided information about a development task, the user-provided information including specifications for the development task and variation of experimental parameters for candidate experiments for the development task;   a module for generating a number of candidate experiments based on the user-provided information about variation of the experimental parameters and calculating preference scores for the candidate experiments based on the user-provided information about specifications for the development task; and   a second user interface for recommending from among the candidate experiments based at least in part on their preference scores.