Intelligence-based medicine: lessons learned after seventy manuscripts

“Maybe that’s enlightenment enough: to know that there is no final resting place of the mind; no moment of smug clarity. Perhaps wisdom… is realizing how small I am, and unwise, and how far I have yet to go.”


-Anthony Bourdaine
American cook/author and global traveler

We have remained in a viral apocalypse now for almost six months, with no obvious denouement to this virtual lockdown. It is not the “new normal” (as some of you may recall, I do not like this term, along with “social distancing”) but a “better normal” that we work adjusting towards.

A special highlight in the past two weeks since I last wrote you was the arrival of the very first copy of the book, Intelligence-Based Medicine that I have had the privilege to edit with many (nearly 100 total) authors of diverse backgrounds contributing commentaries. Many thanks to all those who have supported this effort the past two years (especially all those I failed to thank in the acknowledgment pages). And yes, I am already working on the second edition as this field is moving so very expediently. Besides, I always think the second edition of most books is the best.

Another one of my activities to distract me from this pandemic has been as the privileged reviewer (along with some of you) as well as the editor-in-chief of the many manuscripts that have been submitted to the nascent journal with the same title as the book, Intelligence-Based Medicine (both by Elsevier). After 6 months of reading manuscripts, I have learned that the world remains enthralled with the use of artificial intelligence not only in imaging (by far the biggest category) but also interested in a myriad of other domains (such as cellular automata, multi-agent systems, case-based reasoning, and fuzzy cognitive maps). In addition, submissions were from not only the elite AI in medicine centers but also from about 20 countries from all continents (except Antarctica). This demographic is very much in parallel with our AIMed gatherings (that seem so long ago).

If I may share a few observations (after 70 manuscripts to date) to our readers:

1) People: There could be even more collaboration and synergy between data scientists and clinicians (ones that embrace this dyad are generally stronger papers with higher clinical relevance);

2) Data: The data remain a relatively weak area with studies sometimes working with small, homogenous, inaccurate, and imbalanced data sets as well as data leakage and other issues from improper handling of training, testing, and validation test sets;

3) Data Science: The choice of model was sometimes not optimal and when the model choice is good, the clinical relevance is at times low (the perfect model with zero or little relevance is just as disappointing as a good clinical idea with suboptimal data science).

Thank you all for your support of both the book and the journal as well as AIMed activities.
I absolutely do realize how small I am, how unwise, and how far I have yet to go, but I have also been enlightened early on this journey that there is no resting for my mind and certainly no moment of smug clarity.

Recommended Posts