Summaries of Must-Read Clinical Literature, Guidelines, and FDA Actions
Mining social media to detect adverse drug reactions to newly-approved drugs
Ransohoff J et al. JAMA Oncology March 1, 2018. doi:10.1001/jamaoncol.2017.5688
Key clinical point: DeepHealthMiner, a deep learning pipeline that extracts precise mentions of adverse drug reactions from the informal text in social health networks, may speed up recognition of adverse reactions to new oncology drugs.
Major finding: When the frequency and timing of adverse drug reaction detections were compared, DeepHealthMiner detected adverse cutaneous drug reactions with an average 7-month lead-time from that of clinical reports. In addition, 23 cases of hypohidrosis were detected, constituting a novel drug reaction not previously reported.
Data source: A proof-of-principle comparison of the frequency and timing of adverse cutaneous drug reaction mentions in the Inspire (inspire.com) database and first-published clinical reports of the associations.
Disclosures: The authors reported no conflicts of interest.
Source: Ransohoff J et al. JAMA Oncology March 1, 2018. doi:10.1001/jamaoncol.2017.5688
Ransohoff J et al. JAMA Oncology March 1, 2018. doi:10.1001/jamaoncol.2017.5688