Automatic validation and enrichment of semantic relations between medical entities for drug discovery

1. A method for drug discovery and drug repositioning, implemented in a computer system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor, the method comprising:

generating or adding to a database comprising information relating to symptoms treated by drugs, the information obtained based on data relating to a plurality of aspects of drugs and pharmaceutical compounds;

generating or adding semantic relationships, at the computer system, based on the information in the database relating to symptoms treated by drugs, the generated semantic relationships represented in the form of a semantic graph;

using a Denoising Autoencoder to learn, at the computer system, new patterns, and relationships among the semantic relationships in the semantic graph, wherein the Denoising Autoencoder comprises a deep neural network, and the Denoising Autoencoder is trained to use a hidden layer to generate a particular model based on its inputs; and

using the Denoising Autoencoder to generate, at the computer system, predictions for drug discovery and drug repositioning based on the semantic relationships, including the newly found relations, based on the learned new patterns and relationships.

2. The method of claim 1 , wherein the database is generated or added to by:

collecting, at the computer system, data relating to a plurality of aspects of drugs and pharmaceutical compounds;

extracting, at the computer system, relevant terms from the collected data; and

mapping, at the computer system, the extracted relevant terms to structured medical terms.

3. The method of claim 2 , wherein the semantic relationships are generated or added to by:

generating, at the computer system, semantic relationships represented in the form of a semantic graph based on the mapped structured medical terms.

4. The method of claim 3 , wherein the generated semantic graph comprises nodes and edges between the nodes, the nodes representing entities including at least some of drugs or pharmaceutical compounds, diseases or conditions, and symptoms, and the edges representing relations between the nodes comprising a treats relation and at least some of a causes side effect relation, a has relation, and an indicated relation.

5. The method of claim 4 , wherein the relations between the nodes further comprise a probability of the relation or a score for the relation.

6. The method of claim 5 , wherein the data relating to a plurality of aspects of drugs and pharmaceutical compounds comprises at least some of structured and unstructured data from textual and non-textual sources, including audio sources, video sources, drug labels, medical and drug related databases, medical articles and books, medical health records, social media, internet forums, and tutorials (textual, audio, and video).

7. A system for testing a software system, the system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor to perform:

generating or adding to a database comprising information relating to symptoms treated by drugs, the information obtained based on data relating to a plurality of aspects of drugs and pharmaceutical compounds;

generating or adding semantic relationships, at the computer system, based on the information in the database relating to symptoms treated by drugs, the generated semantic relationships represented in the form of a semantic graph;

using a Denoising Autoencoder to learn, at the computer system, new patterns, and relationships among the semantic relationships in the semantic graph, wherein the Denoising Autoencoder comprises a deep neural network, and the Denoising Autoencoder is trained to use a hidden layer to generate a particular model based on its inputs; and

using the Denoising Autoencoder to generate, at the computer system, predictions for drug discovery and drug repositioning based on the semantic relationships, including the newly found relations, based on the learned new patterns and relationships.

8. The system of claim 7 , wherein the database is generated or added to by:

collecting, at the computer system, data relating to a plurality of aspects of drugs and pharmaceutical compounds;

extracting, at the computer system, relevant terms from the collected data; and

mapping, at the computer system, the extracted relevant terms to structured medical terms.

9. The system of claim 8 , wherein the semantic relationships are generated or added to by:

generating, at the computer system, semantic relationships represented in the form of a semantic graph based on the mapped structured medical terms.

10. The system of claim 9 , wherein the generated semantic graph comprises nodes and vectors between the nodes, the nodes representing entities including at least some of drugs or pharmaceutical compounds, diseases or conditions, and symptoms, and the edges representing relations between the nodes comprising a treats relation and at least some of a causes side effect relation, a has relation, and an indicated relation.

11. The system of claim 10 , wherein the relations between the nodes further comprise a probability of the relation or a score for the relation.

12. The system of claim 11 , wherein the data relating to a plurality of aspects of drugs and pharmaceutical compounds comprises at least some of structured and unstructured data from textual and non-textual sources, including audio sources, video sources, drug labels, medical and drug related databases, medical articles and books, medical health records, social media, internet forums, and tutorials (textual, audio, and video).

13. A computer program product for testing a software system, the computer program product comprising a non-transitory computer readable storage having program instructions embodied therewith, the program instructions executable by a computer, to cause the computer to perform a method comprising:

generating or adding to a database comprising information relating to symptoms treated by drugs, the information obtained based on data relating to a plurality of aspects of drugs and pharmaceutical compounds;

generating or adding semantic relationships, at the computer system, based on the information in the database relating to symptoms treated by drugs, the generated semantic relationships represented in the form of a semantic graph;

using a Denoising Autoencoder to learn, at the computer system, new patterns, and relationships among the semantic relationships in the semantic graph, wherein the Denoising Autoencoder comprises a deep neural network, and the Denoising Autoencoder is trained to use a hidden layer to generate a particular model based on its inputs; and

using the Denoising Autoencoder to generate, at the computer system, predictions for drug discovery and drug repositioning based on the semantic relationships, including the newly found relations, based on the learned new patterns and relationships.

14. The computer program product of claim 13 , wherein the database is generated or added to by:

collecting, at the computer system, data relating to a plurality of aspects of drugs and pharmaceutical compounds;

extracting, at the computer system, relevant terms from the collected data; and

mapping, at the computer system, the extracted relevant terms to structured medical terms.

15. The computer program product of claim 14 , wherein the semantic relationships are generated or added to by:

generating, at the computer system, semantic relationships represented in the form of a semantic graph based on the mapped structured medical terms.

16. The computer program product of claim 15 , wherein the generated semantic graph comprises nodes and vectors between the nodes, the nodes representing entities including at least some of drugs or pharmaceutical compounds, diseases or conditions, and symptoms, and the edges representing relations between the nodes comprising a treats relation and at least some of a causes side effect relation, a has relation, and an indicated relation.

17. The computer program product of claim 16 , wherein the relations between the nodes further comprise a probability of the relation or a score for the relation.

18. The computer program product of claim 17 , wherein the data relating to a plurality of aspects of drugs and pharmaceutical compounds comprises at least some of structured and unstructured data from textual and non-textual sources, including audio sources, video sources, drug labels, medical and drug related databases, medical articles and books, medical health records, social media, internet forums, and tutorials (textual, audio, and video).