Limits...
Automated Detection of Systematic Off-label Drug Use in Free Text of Electronic Medical Records.

Jung K, Lependu P, Shah N - AMIA Jt Summits Transl Sci Proc (2013)

Bottom Line: Studies estimate that 21% of prescriptions are off-label, with only 27% of those uses supported by evidence of safety and efficacy.By including additional features encoding prior knowledge about drugs, diseases, and known usage, we trained a highly accurate predictive model that was used to detect novel candidate off-label usages in a very large clinical corpus.We show that the candidate uses are plausible and can be prioritized for further analysis in terms of safety and efficacy.

View Article: PubMed Central - PubMed

Affiliation: Stanford University, Stanford, CA.

ABSTRACT
Off-label use of a drug occurs when it is used in a manner that deviates from its FDA label. Studies estimate that 21% of prescriptions are off-label, with only 27% of those uses supported by evidence of safety and efficacy. We have developed methods to detect population level off-label usage using computationally efficient annotation of free text from clinical notes to generate features encoding empirical information about drug-disease mentions. By including additional features encoding prior knowledge about drugs, diseases, and known usage, we trained a highly accurate predictive model that was used to detect novel candidate off-label usages in a very large clinical corpus. We show that the candidate uses are plausible and can be prioritized for further analysis in terms of safety and efficacy.

No MeSH data available.


Calculation of drug and disease similarity features. To calculate the similarity of drug m to other drugs used to treat indication n, we find other rows j such that entry (j, n) = 1 (i.e. drug j is known to treat indication n). We calculate the cosine and Jaccard similarities of these rows (indicated by blue arrows) to row m and use the maximum similarity. An analogous calculation is used to calculate indication. similarity.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3814472&req=5

f2-amia_tbi_2013_094: Calculation of drug and disease similarity features. To calculate the similarity of drug m to other drugs used to treat indication n, we find other rows j such that entry (j, n) = 1 (i.e. drug j is known to treat indication n). We calculate the cosine and Jaccard similarities of these rows (indicated by blue arrows) to row m and use the maximum similarity. An analogous calculation is used to calculate indication. similarity.

Mentions: We also used features that encode prior knowledge of the drugs, indications and known usage. These features are motivated by the intuition that drugs are typically used off-label because of similarity with an approved drug, such as a shared molecular target, pathway or drug class — e.g., Bevacizumab, an anti-angiogenic agent, is used off-label to treat age related macular degeneration, a form of blindness caused by aberrant growth of blood vessels (17). We used the Medi-Span and DrugBank databases to construct features that encode this knowledge for each drug-indication pair. For Medi-Span, these included the number of drugs approved for the indication, number of drugs known to be used for the indication, the fraction of known treatments for the indication that are also approved, the maximum similarity of the drug to other drugs known to be used to treat the indication, and the maximum similarity of the indication to other indications treated by the drug. Similarity features were calculated using the cosine and Jaccard similarities as shown in Figure 2 below.


Automated Detection of Systematic Off-label Drug Use in Free Text of Electronic Medical Records.

Jung K, Lependu P, Shah N - AMIA Jt Summits Transl Sci Proc (2013)

Calculation of drug and disease similarity features. To calculate the similarity of drug m to other drugs used to treat indication n, we find other rows j such that entry (j, n) = 1 (i.e. drug j is known to treat indication n). We calculate the cosine and Jaccard similarities of these rows (indicated by blue arrows) to row m and use the maximum similarity. An analogous calculation is used to calculate indication. similarity.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3814472&req=5

f2-amia_tbi_2013_094: Calculation of drug and disease similarity features. To calculate the similarity of drug m to other drugs used to treat indication n, we find other rows j such that entry (j, n) = 1 (i.e. drug j is known to treat indication n). We calculate the cosine and Jaccard similarities of these rows (indicated by blue arrows) to row m and use the maximum similarity. An analogous calculation is used to calculate indication. similarity.
Mentions: We also used features that encode prior knowledge of the drugs, indications and known usage. These features are motivated by the intuition that drugs are typically used off-label because of similarity with an approved drug, such as a shared molecular target, pathway or drug class — e.g., Bevacizumab, an anti-angiogenic agent, is used off-label to treat age related macular degeneration, a form of blindness caused by aberrant growth of blood vessels (17). We used the Medi-Span and DrugBank databases to construct features that encode this knowledge for each drug-indication pair. For Medi-Span, these included the number of drugs approved for the indication, number of drugs known to be used for the indication, the fraction of known treatments for the indication that are also approved, the maximum similarity of the drug to other drugs known to be used to treat the indication, and the maximum similarity of the indication to other indications treated by the drug. Similarity features were calculated using the cosine and Jaccard similarities as shown in Figure 2 below.

Bottom Line: Studies estimate that 21% of prescriptions are off-label, with only 27% of those uses supported by evidence of safety and efficacy.By including additional features encoding prior knowledge about drugs, diseases, and known usage, we trained a highly accurate predictive model that was used to detect novel candidate off-label usages in a very large clinical corpus.We show that the candidate uses are plausible and can be prioritized for further analysis in terms of safety and efficacy.

View Article: PubMed Central - PubMed

Affiliation: Stanford University, Stanford, CA.

ABSTRACT
Off-label use of a drug occurs when it is used in a manner that deviates from its FDA label. Studies estimate that 21% of prescriptions are off-label, with only 27% of those uses supported by evidence of safety and efficacy. We have developed methods to detect population level off-label usage using computationally efficient annotation of free text from clinical notes to generate features encoding empirical information about drug-disease mentions. By including additional features encoding prior knowledge about drugs, diseases, and known usage, we trained a highly accurate predictive model that was used to detect novel candidate off-label usages in a very large clinical corpus. We show that the candidate uses are plausible and can be prioritized for further analysis in terms of safety and efficacy.

No MeSH data available.