Computational platform for doctor-artificial intelligence cooperation in pulmonary arterial hypertension prognostication: a pilot study.
Journal
ERJ open research
ISSN: 2312-0541
Titre abrégé: ERJ Open Res
Pays: England
ID NLM: 101671641
Informations de publication
Date de publication:
Jan 2023
Jan 2023
Historique:
received:
24
09
2022
accepted:
20
10
2022
entrez:
13
2
2023
pubmed:
14
2
2023
medline:
14
2
2023
Statut:
epublish
Résumé
Pulmonary arterial hypertension (PAH) is a heterogeneous and complex pulmonary vascular disease associated with substantial morbidity. Machine-learning algorithms (used in many PAH risk calculators) can combine established parameters with thousands of circulating biomarkers to optimise PAH prognostication, but these approaches do not offer the clinician insight into what parameters drove the prognosis. The approach proposed in this study diverges from other contemporary phenotyping methods by identifying patient-specific parameters driving clinical risk. We trained a random forest algorithm to predict 4-year survival risk in a cohort of 167 adult PAH patients evaluated at Stanford University, with 20% withheld for (internal) validation. Another cohort of 38 patients from Sheffield University were used as a secondary (external) validation. Shapley values, borrowed from game theory, were computed to rank the input parameters based on their importance to the predicted risk score for the entire trained random forest model (global importance) and for an individual patient (local importance). Between the internal and external validation cohorts, the random forest model predicted 4-year risk of death/transplant with sensitivity and specificity of 71.0-100% and 81.0-89.0%, respectively. The model reinforced the importance of established prognostic markers, but also identified novel inflammatory biomarkers that predict risk in some PAH patients. These results stress the need for advancing individualised phenotyping strategies that integrate clinical and biochemical data with outcome. The computational platform presented in this study offers a critical step towards personalised medicine in which a clinician can interpret an algorithm's assessment of an individual patient.
Sections du résumé
Background
UNASSIGNED
Pulmonary arterial hypertension (PAH) is a heterogeneous and complex pulmonary vascular disease associated with substantial morbidity. Machine-learning algorithms (used in many PAH risk calculators) can combine established parameters with thousands of circulating biomarkers to optimise PAH prognostication, but these approaches do not offer the clinician insight into what parameters drove the prognosis. The approach proposed in this study diverges from other contemporary phenotyping methods by identifying patient-specific parameters driving clinical risk.
Methods
UNASSIGNED
We trained a random forest algorithm to predict 4-year survival risk in a cohort of 167 adult PAH patients evaluated at Stanford University, with 20% withheld for (internal) validation. Another cohort of 38 patients from Sheffield University were used as a secondary (external) validation. Shapley values, borrowed from game theory, were computed to rank the input parameters based on their importance to the predicted risk score for the entire trained random forest model (global importance) and for an individual patient (local importance).
Results
UNASSIGNED
Between the internal and external validation cohorts, the random forest model predicted 4-year risk of death/transplant with sensitivity and specificity of 71.0-100% and 81.0-89.0%, respectively. The model reinforced the importance of established prognostic markers, but also identified novel inflammatory biomarkers that predict risk in some PAH patients.
Conclusion
UNASSIGNED
These results stress the need for advancing individualised phenotyping strategies that integrate clinical and biochemical data with outcome. The computational platform presented in this study offers a critical step towards personalised medicine in which a clinician can interpret an algorithm's assessment of an individual patient.
Identifiants
pubmed: 36776484
doi: 10.1183/23120541.00484-2022
pii: 00484-2022
pmc: PMC9907150
pii:
doi:
Types de publication
Journal Article
Langues
eng
Subventions
Organisme : NHLBI NIH HHS
ID : R56 HL152250
Pays : United States
Organisme : NHLBI NIH HHS
ID : R01 HL152250
Pays : United States
Organisme : NHLBI NIH HHS
ID : K25 HL133481
Pays : United States
Organisme : NHLBI NIH HHS
ID : L40 HL134030
Pays : United States
Organisme : NHLBI NIH HHS
ID : K23 HL151892
Pays : United States
Organisme : NHLBI NIH HHS
ID : P01 HL152961
Pays : United States
Informations de copyright
Copyright ©The authors 2023.
Déclaration de conflit d'intérêts
Conflict of interest: None declared.
Références
Nat Mach Intell. 2020 Jan;2(1):56-67
pubmed: 32607472
Clin Chest Med. 2021 Mar;42(1):207-216
pubmed: 33541614
Eur Respir J. 2021 Aug 12;58(2):
pubmed: 33833034
Circulation. 2010 Aug 31;122(9):920-7
pubmed: 20713898
Respir Res. 2014 Apr 16;15:47
pubmed: 24739042
Circ Res. 2014 Jun 20;115(1):165-75
pubmed: 24951765
Am J Respir Crit Care Med. 2021 Jun 15;203(12):1472-1487
pubmed: 33861689
BMC Bioinformatics. 2006 Jan 06;7:3
pubmed: 16398926
Circ Res. 2019 Mar 15;124(6):904-919
pubmed: 30661465
Chest. 2021 Jan;159(1):337-346
pubmed: 32882243
Am J Respir Crit Care Med. 2022 May 1;205(9):1102-1111
pubmed: 35081018
BMC Bioinformatics. 2004 Jun 24;5:81
pubmed: 15217521
BMC Med. 2019 Dec 16;17(1):230
pubmed: 31842878
Eur Respir J. 2015 Oct;46(4):903-75
pubmed: 26318161
Herz. 2019 Apr;44(2):130-137
pubmed: 30847510
Mayo Clin Proc. 2020 Sep;95(9):1978-1988
pubmed: 32861339
Eur Respir J. 2020 Aug 27;56(2):
pubmed: 32366491
EBioMedicine. 2021 Jul;69:103444
pubmed: 34186489
Lancet Respir Med. 2017 Sep;5(9):717-726
pubmed: 28624389
N Engl J Med. 2021 Apr 1;384(13):1204-1215
pubmed: 33789009
Eur Respir J. 2014 Mar;43(3):915-7
pubmed: 24232704
Eur Respir J. 2021 Jun 24;57(6):
pubmed: 33334933
Chest. 2019 Aug;156(2):323-337
pubmed: 30772387