Objective: To evaluate the incorporation of prognostic digital twins to enhance traditional clinical trials in Parkinson’s Disease (PD) for disease modifying therapies. We assess the achievable power gain from incorporating AI-generated digital twins.

Background: We built a generative AI model which can improve traditional clinical trials in PD by enabling more precise decision making. This model forecasts comprehensive health outcomes for participants, which can be used in statistical analyses to increase the precision of treatment effect estimates. In this study, we use 5-fold cross validation over a dataset composed of National Institute of Neurological Disorders and Stroke (NINDS), Parkinson’s Disease Biomarkers Program (PDBP) and Parkinson’s Progression Markers Initiative (PPMI) studies to estimate the benefit of employing digital twins created by our generative AI model in PD trials.

Method: A generative AI model was constructed from data of 2,100+ PD participants from randomized controlled trial control groups and observational studies. The model was provided with each participant’s baseline data and predicted disease progression over a span of 3-18 months across 121 covariates and 91 clinical outcomes. These predictions, or digital twins, are distributions of comprehensive longitudinal trajectories across disease severity outcomes, laboratory measures, and vital signs. We analyzed patients’ scores from parts 1-3 of the Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) score and the MDS-UPDRS subscore. All outcomes were evaluated as changes observed from the initial baseline to the follow-up visit. We compared traditional analysis methods for each endpoint with Bayesian prognostic covariate adjustment, which utilizes the digital twin’s mean and historical data to improve the precision of treatment effect estimation. The achieved variance reduction was translated into increased statistical power.

Results: Digital twins provided benefits across outcomes and time points, boosting power for change in MDS-UPDRS Motor Total Score and Subscore at 1 year from 80% to 87% and 88%, respectively, without enrolling additional participants.

Conclusion: Digital twins can boost power when identifying significant treatment effects in Phase 2 clinical trials of disease modifying therapies for PD. This improvement can be used to reduce risk and bolster the decision making process.

References: [1] Moore CG, Schenkman M, Kohrt WM, Delitto A, Hall DA, Corcos D. Study in Parkinson
disease of exercise (SPARX): translating high-intensity exercise from animals to
humans. Contemp Clin Trials. 2013 Sep;36(1):90-8. doi: 10.1016/j.cct.2013.06.002.
Epub 2013 Jun 14. PMID: 23770108; PMCID: PMC3769494.
[2] Parkinson Study Group SURE-PD Investigators; Schwarzschild MA, Ascherio A, Beal
MF, Cudkowicz ME, Curhan GC, Hare JM, Hooper DC, Kieburtz KD, Macklin EA, Oakes D,
Rudolph A, Shoulson I, Tennis MK, Espay AJ, Gartner M, Hung A, Bwala G, Lenehan R,
Encarnacion E, Ainslie M, Castillo R, Togasaki D, Barles G, Friedman JH, Niles L, Carter
JH, Murray M, Goetz CG, Jaglin J, Ahmed A, Russell DS, Cotto C, Goudreau JL, Russell D,
Parashos SA, Ede P, Saint-Hilaire MH, Thomas CA, James R, Stacy MA, Johnson J,
Gauger L, Antonelle de Marcaida J, Thurlow S, Isaacson SH, Carvajal L, Rao J, Cook M,
Hope-Porche C, McClurg L, Grasso DL, Logan R, Orme C, Ross T, Brocht AF,
Constantinescu R, Sharma S, Venuto C, Weber J, Eaton K. Inosine to increase serum and
cerebrospinal fluid urate in Parkinson disease: a randomized clinical trial. JAMA Neurol.
2014 Feb;71(2):141-50. doi: 10.1001/jamaneurol.2013.5528. PMID: 24366103; PMCID:
PMC3940333.
[3] Parkinson Study Group. DATATOP: a multicenter controlled clinical trial in early
Parkinson’s disease. Arch Neurol. 1989 Oct;46(10):1052-60. doi:
10.1001/archneur.1989.00520460028009. PMID: 2508608.
[4] Parkinson Study Group PRECEPT Investigators. Mixed lineage kinase inhibitor
CEP-1347 fails to delay disability in early Parkinson disease. Neurology. 2007 Oct
9;69(15):1480-90. doi: 10.1212/01.wnl.0000277648.63931.c0. Epub 2007 Sep 19. PMID:
17881719.
[5] Parkinson Study Group. A Controlled Trial of Rasagiline in Early Parkinson Disease:
The TEMPO Study. Arch Neurol. 2002;59(12):1937–1943.
doi:10.1001/archneur.59.12.1937
[6] Shults CW, Oakes D, Kieburtz K, Beal MF, Haas R, Plumb S, Juncos JL, Nutt J,
Shoulson I, Carter J, Kompoliti K, Perlmutter JS, Reich S, Stern M, Watts RL, Kurlan R,
Molho E, Harrison M, Lew M; Parkinson Study Group. Effects of coenzyme Q10 in early
Parkinson disease: evidence of slowing of the functional decline. Arch Neurol. 2002
Oct;59(10):1541-50. doi: 10.1001/archneur.59.10.1541. PMID: 12374491.
[7] Fahn S, Oakes D, Shoulson I, Kieburtz K, Rudolph A, Lang A, Olanow CW, Tanner C,
Marek K; Parkinson Study Group. Levodopa and the progression of Parkinson’s disease.
N Engl J Med. 2004 Dec 9;351(24):2498-508. doi: 10.1056/NEJMoa033447. PMID:
15590952.
[8] Wills AM, Li R, Pérez A, Ren X, Boyd J; NINDS NET-PD Investigators. Predictors of
weight loss in early treated Parkinson’s disease from the NET-PD LS-1 cohort. J Neurol.
2017 Aug;264(8):1746-1753. doi: 10.1007/s00415-017-8562-4. Epub 2017 Jul 15. PMID:
28712000; PMCID: PMC5789795.
[9] Parashos SA, Luo S, Biglan KM, Bodis-Wollner I, He B, Liang GS, Ross GW, Tilley BC,
Shulman LM; NET-PD Investigators. Measuring disease progression in early Parkinson
disease: the National Institutes of Health Exploratory Trials in Parkinson Disease
(NET-PD) experience. JAMA Neurol. 2014 Jun;71(6):710-6. doi:
10.1001/jamaneurol.2014.391. PMID: 24711047; PMCID: PMC4188544.
[10] NINDS Exploratory Trials in Parkinson Disease (NET-PD) FS-ZONE Investigators.
Pioglitazone in early Parkinson’s disease: a phase 2, multicentre, double-blind,
randomised trial. Lancet Neurol. 2015 Aug;14(8):795-803. doi:
10.1016/S1474-4422(15)00144-1. Epub 2015 Jun 23. Erratum in: Lancet Neurol. 2015
Oct; 14(10):979. PMID: 26116315; PMCID: PMC4574625.
*PPMI data version 28Nov2023

« Back to 2024 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/boosting-clinical-trial-power-in-parkinsons-disease-with-ai-generated-digital-twins/