DeLaffolie J.: Artificial Intelligence

Today’s ESPGHAN podcast interviews Dr Jan de Laffolie, at present in Giessen (Germany), who has a strong interest in both inflammatory bowel disease and in “artificial intelligence”, or AI – that is, the sifting of data for correlations, not only those that are apparent to humans unassisted, call it to the naked eye / the naked mind, but also those that elude us mortals when we are on our own.  Working within a tripartite framework of :  

1) What is artificial intelligence and how can it be applied to paediatric gastroenterology, hepatology, and nutrition (PGHN) ?

2) What are disadvantages and risks associated with developing AI in medicine in general and in PGHN in particular ?

3) What do patients and families need to know about data security and AI ?

Dr de Laffolie presents a brief précis of how AI developed, pari passu with increased computational power, moving from “expert systems” with direct interrogation of a database through multilayered systems, in which filtration occurs in a database between question and answer in processes masked from the questioner, and onward to “deep learning” that integrates more than one database into the filtration.  AI already is used, he says, in surveillance endoscopy, improving (in adults) adenoma detection by a factor of 33%.  It also is of value in infectious-disease management, adult and paediatric alike, permitting those attending an infected patient to choose a treatment regimen on the basis of others’ concrete experience rather than that of theories.  But it must be used with care :  In hoping to profit by AI, patients or their parents confer on, confide in !, various databases information that might prove, somewhere down the road, even years later, disadvantageous to the propositus.  This possibility must be explained, with its long “lead time”, to the releasers of information to allow, to the extent that this is possible, informed consent.  Within these bounds AI can be, he believes, not a bad master that is held up for us to fear but instead a good servant.

Literature :

Schneider N et al.  Machine learning classification of inflammatory bowel disease in children based on a large real-world pediatric cohort CEDATA-GPGE® Registry.  Front Med (Lausanne) 2021 May 24:8:666190.  doi: 10.3389/fmed.2021.666190.  eCollection 2021.  PMID: 34109197

Brooks-Warburton J et al.  Frontline Gastroenterol 2021 Dec 1013(4):325-331.  Artificial intelligence and inflammatory bowel disease:  Practicalities and future prospects.  doi: 10.1136/flgastro-2021-102003.  eCollection 2022.  PMID : 35722596

Stidham RW, Takenaka K.  Artificial intelligence for disease assessment in inflammatory bowel disease:  How will it change our practice?  Gastroenterology 2022 Apr 162(5):1493-1506.  doi: 10.1053/j.gastro.2021.12.238.  Epub 2022 Jan 4.  PMID :  34995537

Dr. DeLaffolie´s favourite song: Die perfekte Welle - Julie https://open.spotify.com/track/2LV5joNDrsyuXEh4FBARVK?si=d12211db83c74180 

ESPGHAN favourite Songs can be found on Spotify https://open.spotify.com/playlist/0YIHKjxITLEm9XNyHyypTo