PMID- 38219135
OWN - NLM
STAT- MEDLINE
DCOM- 20240319
LR  - 20240319
IS  - 1526-3231 (Electronic)
IS  - 0749-8063 (Linking)
VI  - 40
IP  - 4
DP  - 2024 Apr
TI  - Editorial Commentary: Artificial Intelligence Analysis of Biomedical, Large, 
      Clinical Registry Data Using Machine Learning Requires Tens of Thousands of 
      Subjects and a Focus on Substantial Clinical Benefit: Minimal Clinically 
      Important Difference Is too Low a Bar.
PG  - 1164-1167
LID - S0749-8063(23)00879-4 [pii]
LID - 10.1016/j.arthro.2023.10.035 [doi]
AB  - As the field of arthroscopic hip preservation surgery grows, large high-quality 
      registries represent a foundational study design for establishing whether hip 
      arthroscopy is effective for patients with femoroacetabular impingement syndrome 
      (FAIS). Original research publications from experienced high-volume surgeons tell 
      us "Can it work". A registry tells us "Does it work?". The ability of 
      preservation to truly preserve the joint, delay the arthritis process, and reduce 
      the risk of arthroplasty requires long-term follow-up. A geographic registry can 
      follow this. The registry represents the "real world", a heterogeneous set of 
      variables pertaining to the doctor, patient, intervention, and outcome. The vast 
      array of factors that can be analyzed before, during, and after surgery makes 
      machine learning an ideal technique for analysis of large quantities of data. A 
      global hip preservation surgery registry is a desirable and achievable goal. In 
      order to optimally predict outcome of hip arthroscopy, given the known large 
      number of patient- and hip-specific factors that influence outcomes, a deep 
      learning model with tens of thousands of subjects for this medium-scale task 
      would be needed. Measures of clinical relevance need to include more than just 
      MCID (minimal clinically important difference), which is the lowest bar minimal 
      threshold. Patient expectations often far exceed MCID-requiring other metrics 
      like SCB (substantial clinical benefit), PASS (patient acceptable symptom state), 
      and MOI (maximal outcome improvement). Registries should include validated, 
      reliable, and responsive patient-reported outcome scores (e.g., International Hip 
      Outcome Tool [iHOT-12]) with measures of clinical relevance and expectations 
      assessed routinely. The United Kingdom's NAHR (Non-Arthroplasty Hip Registry) and 
      Denmark's DHAR (Danish Hip Arthroscopy Registry) are the two largest 
      geography-based registries in current hip preservation research both with 11 
      years of patient enrollment.
CI  - Copyright (c) 2023 Arthroscopy Association of North America. Published by Elsevier 
      Inc. All rights reserved.
FAU - Harris, Joshua D
AU  - Harris JD
LA  - eng
PT  - Editorial
DEP - 20240111
PL  - United States
TA  - Arthroscopy
JT  - Arthroscopy : the journal of arthroscopic & related surgery : official 
      publication of the Arthroscopy Association of North America and the International 
      Arthroscopy Association
JID - 8506498
SB  - IM
MH  - Humans
MH  - Treatment Outcome
MH  - Artificial Intelligence
MH  - Minimal Clinically Important Difference
MH  - Routinely Collected Health Data
MH  - Activities of Daily Living
MH  - *Femoracetabular Impingement/surgery
MH  - Machine Learning
MH  - Arthroscopy/methods
MH  - Registries
MH  - Patient Reported Outcome Measures
MH  - *Arthroplasty, Replacement
MH  - Hip Joint/surgery
MH  - Retrospective Studies
MH  - Follow-Up Studies
COIS- Disclosure The authors report the following potential conflicts of interest or 
      sources of funding: J.D.H. reports consulting and advisory fees relationship with 
      Smith and Nephew Inc.; consulting and advisory fees from Slack Incorporated; 
      consulting and advisory fees from Thieme Medical Publishers; board membership in 
      International Society of Arthroscopy Knee Surgery and Orthopaedic Sports Medicine 
      and Arthroscopy Association of North America; and editorial board membership of 
      Arthroscopy. Full ICMJE author disclosure forms are available for this article 
      online, as supplementary material.
EDAT- 2024/01/14 14:42
MHDA- 2024/03/19 06:43
CRDT- 2024/01/14 10:34
PHST- 2023/10/22 00:00 [received]
PHST- 2023/10/26 00:00 [accepted]
PHST- 2024/03/19 06:43 [medline]
PHST- 2024/01/14 14:42 [pubmed]
PHST- 2024/01/14 10:34 [entrez]
AID - S0749-8063(23)00879-4 [pii]
AID - 10.1016/j.arthro.2023.10.035 [doi]
PST - ppublish
SO  - Arthroscopy. 2024 Apr;40(4):1164-1167. doi: 10.1016/j.arthro.2023.10.035. Epub 
      2024 Jan 11.