Machine learning techniques for automatic evaluation of clinical trial data
Aspects of the subject matter described in this specification are embodied in systems and methods that utilize machine-learning techniques to evaluate clinical trial data using one or more learning models trained to identify anomalies representing adverse events associated with a clinical trial investigation. In some implementations, investigation data collected at a clinical trial site is obtained. A set of models corresponding to the clinical trial site is selected. Each model included in the set of models is trained to identify, based on historical investigation data collected at the clinical trial site, a distinct set of one or more indicators that indicate a compliance risk associated with the investigation data. A score for the clinical trial site is determined based on the investigation data relative to the historical investigation data. The score represents a likelihood that the investigation data is associated with at least one indicator representing the compliance risk.
1 . A computer-implemented method comprising:
obtaining, from a plurality of data sources, clinical trial investigation data comprising a set of medical records, wherein the set of medical records comprise data from a first set of records represented in a first data format and data from a second set of records represented in a second data format, the first data format and the second data format being incompatible with one another;
converting the clinical trial investigation data into a standardized data format that associates corresponding data fields in the first data format and the second data format;
selecting, based on one or more lookup parameters associated with the set of medical records of a clinical trial, a set of indicators for evaluation of the set of medical records, wherein each indicator included in the set of indicators specifies a different condition representing a likely anomaly for a medical record included in the set of medical records;
recognizing, based on querying, by a trained machine learning model, the clinical trial investigation data in the standardized data format, a subset of medical records that satisfy at least one of the conditions specified by the set of indicators;
determining, using the trained model, a score for each medical record included in the subset of medical records; and
generating, based on the score, a report that indicates whether the investigation data represents an adverse event.
2 . The computer-implemented method of claim 1 , further comprising organizing the clinical trial investigation data in the standardized format into a hierarchical data schema that stores investigator-level records at a lower hierarchical level and site-level records at a higher hierarchical level.
3 . The computer-implemented method of claim 1 , wherein the one or more lookup parameters comprises a lookup parameter indicating a medication associated with the clinical trial.
4 . The computer-implemented method of claim 1 , wherein the set of indicators comprises an indicator specifying an expected dosage for a medication.
5 . The computer-implemented method of claim 4 , wherein the subset of medical records includes a medical record associated with a patient that is identified by the trained model as being prescribed a dosage of the medication that exceeds the expected dosage for the medication.
6 . The computer-implemented method of claim 1 , wherein the one or more lookup parameters comprise a lookup parameter indicating a disease condition associated with the clinical trial.
7 . The computer-implemented method of claim 6 , wherein the set of indicators comprises an indicator specifying an expected treatment plan for the disease condition.
8 . The computer-implemented method of claim 7 , wherein the subset of medical records includes a medical record associated with a treatment event that is identified by the trained model as being deviating from the expected treatment plan for the disease condition.
9 . A system comprising:
one or more computers and one or more devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
obtaining, from a plurality of data sources, clinical trial investigation data comprising a set of medical records, wherein the set of medical records comprise data from a first set of records represented in a first data format and data from a second set of records represented in a second data format, the first data format and the second data format being incompatible with one another;
converting the clinical trial investigation data into a standardized data format that associates corresponding data fields in the first data format and the second data format;
selecting, based on one or more lookup parameters associated with the set of medical records of a clinical trial, a set of indicators for evaluation of the set of medical records, wherein each indicator included in the set of indicators specifies a different condition representing a likely anomaly for a medical record included in the set of medical records;
recognizing, based on querying, by a trained machine learning model, the clinical trial investigation data in the standardized data format, a subset of medical records that satisfy at least one of the conditions specified by the set of indicators;
determining, using the trained model, a score for each medical record included in the subset of medical records; and
generating, based on the score, a report that indicates whether the investigation data represents an adverse event.
10 . The system of claim 9 , further comprising organizing the clinical trial investigation data in the standardized format into a hierarchical data schema that stores investigator-level records at a lower hierarchical level and site-level records ad a higher hierarchical level.
11 . The system of claim 9 , wherein the one or more lookup parameters comprises a lookup parameter indicating a medication associated with the clinical trial.
12 . The system of claim 9 , wherein the set of indicators comprises an indicator specifying an expected dosage for a medication.
13 . The system of claim 12 , wherein the subset of medical records includes a medical record associated with a patient that is identified by the trained model as being prescribed a dosage of the medication that exceeds the expected dosage for the medication.
14 . The system of claim 9 , wherein the one or more lookup parameters comprise a lookup parameter indicating a disease condition associated with the clinical trial.
15 . The system of claim 14 , wherein the set of indicators comprises an indicator specifying an expected treatment plan for the disease condition.
16 . The system of claim 15 , wherein the subset of medical records includes a medical record associated with a treatment event that is identified by the trained model as being deviating from the expected treatment plan for the disease condition.
17 . A non-transitory computer-readable medium storing software comprising instructions executable by one or more computers which, when executed, cause the one or more computers to perform operations comprising:
obtaining, from a plurality of data sources, clinical trial investigation data comprising a set of medical records, wherein the set of medical records comprise data from a first set of records represented in a first data format and data from a second set of records represented in a second data format, the first data format and the second data format being incompatible with one another;
converting the clinical trial investigation data into a standardized data format that associates corresponding data fields in the first data format and the second data format;
selecting, based on one or more lookup parameters associated with the set of medical records of a clinical trial, a set of indicators for evaluation of the set of medical records, wherein each indicator included in the set of indicators specifies a different condition representing a likely anomaly for a medical record included in the set of medical records;
recognizing, based on querying, by a trained machine learning model, the clinical trial investigation data in the standardized data format, a subset of medical records that satisfy at least one of the conditions specified by the set of indicators;
determining, using the trained model, a score for each medical record included in the subset of medical records; and
generating, based on the score, a report that indicates whether the investigation data represents an adverse event.
18 . The non-transitory computer-readable medium of claim 17 , further comprising organizing the clinical trial investigation data in the standardized format into a hierarchical data schema that stores investigator-level records at a lower hierarchical level and site-level records at a higher hierarchical level.
19 . The non-transitory computer-readable medium of claim 17 , wherein the one or more lookup parameters comprises a lookup parameter indicating a medication associated with the clinical trial.
20 . The non-transitory computer-readable medium of claim 17 , wherein the set of indicators comprises an indicator specifying an expected dosage for a medication.