TY - JOUR
T1 - Visible Machine Learning for Biomedicine
AU - Yu, Michael K.
AU - Ma, Jianzhu
AU - Fisher, Jasmin
AU - Kreisberg, Jason F.
AU - Raphael, Benjamin J.
AU - Ideker, Trey
N1 - Funding Information:
We are indebted to many individuals for conversations that inspired and informed this commentary, including Terry Sejnowski, Andrea Califano, Michael Kramer, and Janusz Dutkowski. We are grateful for the help of Aidan Ideker, Cherie Ng, and Charlotte Curtis in building the Visible V8 engine model shown in Figure 2A. This work was supported by grants from the NIH ( R01 HG009979 , OT3 TR002026 , and P41 GM103504 to T.I. and R01 HG007069 and U24 CA211000 to B.J.R.).
Publisher Copyright:
© 2018
PY - 2018/6/14
Y1 - 2018/6/14
N2 - A major ambition of artificial intelligence lies in translating patient data to successful therapies. Machine learning models face particular challenges in biomedicine, however, including handling of extreme data heterogeneity and lack of mechanistic insight into predictions. Here, we argue for “visible” approaches that guide model structure with experimental biology. A major ambition of artificial intelligence lies in translating patient data to successful therapies. Machine learning models face particular challenges in biomedicine, however, including handling of extreme data heterogeneity and lack of mechanistic insight into predictions. Here, we argue for “visible” approaches that guide model structure with experimental biology.
AB - A major ambition of artificial intelligence lies in translating patient data to successful therapies. Machine learning models face particular challenges in biomedicine, however, including handling of extreme data heterogeneity and lack of mechanistic insight into predictions. Here, we argue for “visible” approaches that guide model structure with experimental biology. A major ambition of artificial intelligence lies in translating patient data to successful therapies. Machine learning models face particular challenges in biomedicine, however, including handling of extreme data heterogeneity and lack of mechanistic insight into predictions. Here, we argue for “visible” approaches that guide model structure with experimental biology.
UR - http://www.scopus.com/inward/record.url?scp=85048162030&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85048162030&partnerID=8YFLogxK
U2 - 10.1016/j.cell.2018.05.056
DO - 10.1016/j.cell.2018.05.056
M3 - Comment/debate
C2 - 29906441
AN - SCOPUS:85048162030
SN - 0092-8674
VL - 173
SP - 1562
EP - 1565
JO - Cell
JF - Cell
IS - 7
ER -