Data Science and Machine Learning
Data ridden systems are a frequent occurrence in all fields of human
endeavor as computers have the means to store an incredible amount of
data. Biological engineering, of particular interest to us, is no
exception in this regard as advances in measurement techniques have
helped collect massive amounts of data in numerous areas in which
chemical engineers are active. Venkatasubramanian provides an
interesting perspective of artificial intelligence in chemical
engineering.81 He is cautiously optimistic, and
observes that data science and machine learning with suitable infusion
of first principles could be an attractive combination for chemical
engineering applications. Diagnostic issues in health science are indeed
a promising area of application. Verma, in quest of the source of
peripheral neuropathy from the use of Vincristine in the treatment of
leukemic cancer, used machine learning to identify a handful of pain
associated metabolites.82 Early detection of such
metabolites from blood samples could lead to drug dosage adjustments for
improving the quality of life.
Lumping of chemical species in systems with a very large number of
species has been of interest to chemical reaction engineers. When
nonlinear reaction kinetics is involved, lumping strategies could be of
interest using data science and machine learning. Similarly, order
parameters or collective variables have been of interest in molecular
systems towards reducing their dimension as a means to study rare events
such as nucleation. Data science and machine learning could be a
potential source to elucidate a small number of collective variables in
terms of which the free energy of the system can be determined. While
such applications abound, the likelihood of pedestrian usage of the
methodology cannot be ignored.