Journal of Surgery (ISSN: 2575-9760)

short commentary

Comments for Unintended Consequences of Machine Learning in Medicine

Zhong Jia1*, Jie Zhang1, Chao-Jun Kong2

 1Hangzhou First People's Hospital, Nanjing Medical University Affiliated Hangzhou Hospital, Hangzhou, Zhejiang, China

2Chinese Medicine University Fourth Affiliated Clinical Hospital, Hangzhou, Zhejiang,China

 *Corresponding author: Zhong Jia, Department of Hepatopancreatobililiary Surgery, Huansha Road No.261, Hangzhou First People's Hospital, Nanjing Medical University Affiliated Hangzhou Hospital, Hangzhou, Zhejiang 310006, China. Tel: +8613958114181; Fax: +86057187914773; Email:

Received Date: 01 October, 2017; Accepted Date: 02 October, 2017; Published Date: 09 October, 2017


In the era of big data-based “Artificial intelligence”, machine learning is becoming a key core of application technologies now and future. But in the course of its growth and improvement, many negative factors,including the incomplete clinical data, lacking of optimal algorithm,etc. may take inaccurate or even counterproductive effects on ML-DSS,so it's not necessary to make a fuss.It indeed is just a fike if someone has over worrisome attitude about the new emerging advance [1] in this article would like to express their real concerns, aiming to transmit alert messages to first-line clinicians as overreliance of automated results may accidently produce unintended consequences, such as weakening clinical skills, decision-making accuracy of the machine, etc. It also has raised upset in the readers.The authors' starting point is good, with intention to alert clinicians not ignore unintended consequence of machine. But this is just begging for sick days. In fact, machine brain is to extend human brain rather than replace human brain. Sometimes, our experience and inertia will also mislead to accidental errors or results. In practice, we usually recheck results from automated machine by manual review, so the final judgment is determined by clinicians particularly regarding to the critical value so as to remind and register on record book, and then to urge clinicians to take steps. Labor tools let hands free, while deepmachine learning liberates human brain. But when machine thinks like a human, we are really worry about due to its cool metal without any emotions.Just like exploration of nuclear energy, machine learning in medical fields must be under control and have essential constraints in medicine, if any.