Sage Journals: Discover world-class research

Abstract

Recent years have seen successes in symptomatic drugs for Parkinson's disease, but the development of treatments for stopping disease progression continues to fail in clinical drug trials, largely due to the lack of clinical efficacy of drugs. This may be related to limited understanding of disease mechanisms, data heterogeneity, poor target screening and candidate selection, challenges in determining optimal dosage levels, reliance on animal models, insufficient patient participation, and lack of drug adherence in trials. Most of the recent applications of digital health technologies and artificial intelligence (AI)-based tools focused mainly on stages before clinical drug trials. Recent applications used AI-based algorithms or models to discover novel targets, inhibitors and indications, recommend drug candidates and drug dosage, and promote remote data collection. This paper reviews the state of the literature and highlights strengths and limitations in digital approaches to drug discovery and development for Parkinson's disease from 2021 to 2024, and offers recommendations for future research and practice for the success of drug clinical trials.

Plain language summary

In recent years, there have been successes in developing drugs that can help people with Parkinson's disease manage their symptoms but developing drugs that can actually stop the disease from getting worse has been a continuing challenge. Artificial intelligence (AI) has shown promise in improving the success of drug discovery and development for Parkinson's disease. This paper provided an overview of the existing challenges to successful drug trials for Parkinson's disease, recent digital approaches to improve trial success, and remaining challenges that need to be addressed in the future. We found that digital health technologies and AI-based tools were useful for improving the success of clinical drug trials by finding new drug targets and suitable candidates, improving efficiency in pre-clinical evaluation, and facilitating more convenient drug monitoring. While they are helpful in driving innovation and efficiency, drug clinical trials still continue to fail. This may be related to several factors, such as a limited understanding of disease mechanisms, a lack of animal models that can mimic Parkinson's disease and limited real-world patient data for algorithm training. The use of these technologies in clinical drug trials is in its early stages. More research is still needed better to understand the root causes of Parkinson's disease. Researchers are also encouraged to explore how to use existing digital tools to help with drug development at later stages, like large-scale clinical trials and monitoring of drugs after they are approved.

Keywords

Parkinson's disease drug discovery clinical trial technology

Background

Over the past 15–20 years, digital health technologies (DHTs) and artificial intelligence (AI) have been applied across various stages of drug research and development, from target discovery to post-market monitoring.^1,2 DHTs are systems that use computing platforms, connectivity, software, and/or sensors for healthcare and related uses.³ These systems can support patient-centered approaches to drug development with improved measurement of disease progression, such as remote data collection.^4,5 With the power of AI—the ability of algorithms to learn from data and perform tasks automatically without explicit human programming⁶—these systems have offered opportunities to accelerate the drug discovery and development process.^7,8

The drug development process involves rigorous testing through several phases: Pre-clinical trials assess a candidate's safety in labs and animal models. This is followed by Phase I trials to determine the safety and optimal dosage in small groups of human participants. Phase II then evaluates the drug's efficacy in slightly larger groups of patients. Phase III involves large populations to confirm the drug's effectiveness and safety for regulatory approval. Upon drug approval, the side effects, effectiveness and/or cost will be monitored in Phase IV.⁹ 90% of the drug candidates that passed preclinical evaluation failed during clinical trials and approval.¹⁰ For neurodegenerative diseases like Parkinson's disease (PD), drug development is even more challenging due to the complexities of the diseases.⁷ PD is a gradually progressive neurodegenerative condition.¹⁰ It is commonly treated with symptomatic drugs that alleviate the motor and nonmotor symptoms, such as levodopa, dopamine agonists and monoamine oxidase-B inhibitors.¹¹ There is still no established disease-modifying treatment that can slow, stop or reverse the progression of PD.¹² Many attempts were made over the years in this regard, but efforts to translate promising preclinical findings into successful clinical outcomes have continued to fall short.^14–16

The low success rate in clinical trials of disease-modifying treatments for PD can be attributed to several factors.^10,17 Preclinical challenges include difficulties in identifying potential drug targets due to limited understanding of the underlying disease mechanisms, lack of sufficient representative in-vitro models (i.e., experiments conducted outside of a living organism) for selecting promising drug candidates,¹⁷ lack of in vivo models (i.e., experiments conducted within living organisms) that accurately recapitulate the pathology of PD,¹⁸ as well as issues with determining optimal dosage levels,¹⁹ and lack of suitable cellular phenotypes for accurate drug screening.²⁰ Failures in clinical trials are mainly due to a lack of clinical efficacy and unmanageable toxicity.¹⁰ Lack of patient participation in trials may also cause delays in obtaining results, small sample size, and shorter duration of experiments, further limiting the generalizability of findings.^21,22

Previous reviews have presented the potential and challenges in the application of digital technologies and AI for various applications. These include drug discovery in central nervous system diseases,²³ general drug repurposing/repositioning,²⁴ medical devices, pharmaceutical care, and biotechnology,² and diagnosis and management of PD.²⁵ Most of them reviewed studies before 2021. In light of the rapid advancements in technologies, this paper offers a review of the current state of technologies, major challenges, and future outlook for drug design and development for PD in this rapidly evolving field. This paper will particularly explore research on DHTs and AI within the past three years. Drawing insights from previous clinical trials, this review will offer recommendations to address limitations and improve study designs for the success of clinical drug trials for PD.

Digital approaches to drug discovery and development

Recent years have seen advances in the applications of novel targets and inhibitors, drug repurposing, drug candidate selection, patient participation and drug monitoring for PD (see Table 1 and Figure 1). This section introduces the digital approaches to drug discovery and development in these key areas of applications.

Figure 1.

Take-home summary.

Table 1.

Summary of failures and digital approaches to drug discovery and development.

Application field	Past failure	Technology used	Contribution	Example
Target identification	Failure to demonstrate effectiveness for indications	Classification algorithms	Identification of patterns in existing patient data to predict disease-target associations and inform personalized treatment	Lin et al., 2023²⁶
Target identification	Failure to demonstrate effectiveness for indications	Machine learning algorithms like LASSO and mSVM-RFE	Identification of genes associated with immune cell types for PD, enhancing understanding about mechanisms underlying PD and helping discover new inhibitors	Di Stefano et al., 2022²⁷
Drug repurposing and candidate selection	Phase III trials (where drug is tested for effectiveness in large groups)	Diffusion algorithms	Recommending potential drugs for repurposing based on similarity to known effective drugs, to improve likelihood of effectiveness	Chen et al., 2021¹⁸
Drug repurposing and candidate selection		Classification algorithms	Identification of associations between existing drugs and diseases, aiding discovery of new repurposing opportunities	Liu et al., 2022²⁸
Drug candidate selection	Lack of in vivo animal models that fully mirror PD pathology	Human motor system-based robot-on-a-chips	More accurate representation of human muscle movement than traditional organoids-on-a-chip	Shin et al., 2024²⁹
Patient participation and drug monitoring	Non-adherence to drugs during clinical trials	DHTs such as mobile apps and wearable sensors	Monitoring medication adherence in real-time and reminding patients to take the trugs	Christie et al., 2021³⁰; Moreau et al., 2023³¹

Novel targets and inhibitors

A key reason for clinical drug trial failures is lack of efficacy, where drugs fail to demonstrate effectiveness in their indications. AI-based approaches have been widely applied to identify new drug targets and inhibitors, aiming to improve the specificity and efficacy of treatments for PD. A drug target is the biomolecule a drug interacts with (usually specific to a disease), and an inhibitor is the agent that reduces the target's activity, which helps to demonstrate the target's role in disease mechanisms.³³ By directly targeting disease-specific pathways or molecules, drugs can be more effective in treating the condition. Machine learning methods have been used to support the discovery of drug targets and new inhibitors, such as identifying neuroprotective hexapeptides from walnut protein,²⁷ identifying the most important genes associated with immune cell types in the brain for PD,³⁴ discovering novel potential inhibitors for cyclin-dependent kinases 5 (Cdk5) through virtual screening,²⁸ and identifying potent protein aggregation inhibitors for optimization of in vitro assays.³⁵

While AI can help to identify potentially disease-associated genes, the functional validation of these genes remains challenging. For neurodegenerative diseases, such as Alzheimer's disease, it is difficult to validate AI-predicted targets due to the lack of highly reproducible experimental models.³⁵ Additionally, AI models often detect correlations rather than causation, leading to the selection of targets that are associated with PD but may not play a direct mechanistic role in disease progression. For example, variants in the Alpha-synuclein (SNCA) gene are the major genes influencing PD, but the association between an SNCA polymorphism and individual outcome still remains unclear.³⁶ Likewise, lack of understanding of target engagement also contributed to failures in clinical trials targeting α-synuclein in people with PD.³⁷ The k-nearest neighbors algorithm was used to predict drug complexities for PD, but its ability to model complex drug-disease interactions is still limited.³⁸

To bridge this gap, AI-based multi-omics analysis can be used to provide insights into the molecular mechanisms underlying PD.³⁹ Integrating causal inference models may also facilitate understanding the causal relationship between specific drug targets and disease mechanisms. A recent causal-enhanced method demonstrated higher accuracy in predicting drug–target interaction than traditional AI methods by integrating graph generation and multi-source information fusion.⁴⁰ Causal AI, a predictive AI technology, may also be applied to identify cause-and-effect dependencies between variables in drug discovery.⁴¹ Such methods may be promising for advancing target discovery for PD.

Drug repurposing and candidate selection

Drug development is a lengthy and costly process.^23,42 Drug repurposing has emerged as a promising strategy to accelerate drug development by identifying new indications for existing drugs, saving 3–12 years in development time and cost.⁴³ ML-driven knowledge graphs can be used to model the interaction probability between drugs and diseases, facilitating the identification of novel drug candidates and indications.⁴⁴ Diffusion algorithms (e.g., graph diffusion algorithm) were used to recommend potential drugs based on their similarity to known effective drugs or their predicted interactions with disease-related targets.¹⁸ Classification algorithms have also been used to identify potential indications for existing drugs based on their associations with PD-related proteins (e.g., LProts),²⁸ or associating PD genes with Alzheimer's disease.⁴⁵ Natural language processing-based knowledge graph was also used for identifying the most promising drug candidates,⁴⁶ and a combination of deep neural network and ImageNet were adopted for disease modeling and drug screening.²⁰

While AI has accelerated the identification of repurposing candidates, AI-recommended drugs still have not seen success in clinical trials. Using IBM Watson, acamprosate, ganaxolone, and lorcaserin were identified as lead candidate drugs with the potential for repurposing to treat L-DOPA-induced dyskinesia, but none ultimately proved successful in in vivo studies.⁴⁷ In another example, researchers used an AI-based drug repurposing platform (Standigm Insight™) and identified efavirenz as a modulator of α-synuclein propagation; while it showed potential in animal models, its therapeutic efficacy still requires further investigation.⁴⁸

The failure of AI-driven drug repurposing can be attributed to several factors. One key reason is the discrepancies between animal models and human responses. Animal models are used in vivo models and are paramount to test drug candidates in the stage of preclinical evaluation to assess their safety, efficacy and toxicity before human trials.⁴⁹ Currently, animal models used for testing could not fully recapitulate PD pathology, causing failures in clinical drug trials.^18,44,50 AI models make predictions based on existing data.⁵¹ If data are from animal models, this may lead to potential false positives in AI predictions. A breakthrough in this regard is the use of human motor system-based robot-on-a-chips for evaluating drug effects on neurodegenerative diseases, particularly PD.²⁹ This advance can contribute to advancement in preclinical drug evaluation, but its benefits beyond preclinical screening are still unclear.

Another big challenge in applying AI algorithms to clinical settings is the limited availability and adoption of real-world, high-quality data (e.g., electronic health records) for model training.²⁵ The distribution differences between training data and real-world data (also known as data shifts) could lead to a decline in a clinical AI system's performance.⁵² This may explain why models that perform well in simulations or on curated datasets still fail to generalize to the diverse and complex patient populations encountered in actual clinical trials. Improving AI models’ resistance to data shifts may contribute to narrowing the gap between model training and real-world applications. This could be achieved by deliberately training the model with noisy data and early evaluation of dataset shifts.⁵³ Strategies such as synthetic data generation,⁵⁴ transfer learning,⁵⁵ and federated learning⁷ unifying frameworks for dataset shift diagnostics (e.g.,^56,57) may enhance the generalization of results from one setting to another, so as to improve the clinical translation of AI-powered solutions for PD.

DHTs for greater patient participation and drug monitoring

In addition to the above issues, obstacles to clinical drug trials for PD also include insufficient patient participation and a lack of high-quality data from diverse patient populations.⁵⁸ These may cause delays in trial completion, enlarge disparities, and limit the generalizability of study results.²¹ Patients with PD often have difficulties in walking and other motor functions, which prevents them from participating in clinical trials in person.

Digital health tools (e.g., mobile apps and wearable sensors) can offer remote monitoring of PD symptoms, thereby promoting patient participation in drug trials.^59,60 AI-powered wearable sensors, such as Parkinson's KinetiGraph (PKG),⁶¹ allow continuous tracking of motor symptoms, while smartphone applications like Roche PD Mobile Application v2 collect data on bradykinesia, bradyphrenia, speech, tremor, gait and balance.⁶² DHTs have also been used to address patients’ non-adherence to prescribed drugs during clinical trials, which represents an ongoing challenge to the efficacy evaluation of drugs.³⁰ For example, “Smart” pill packs were used to wirelessly track pill usage and send data back to the research team, enhancing drug monitoring and providing data for further drug development.³⁰ Another example is Stat-On^TM, which integrates movement monitoring and drug intake recording to help patients track their symptoms and improve drug adherence.³¹ Moreover, DHTs have the potential to enhance patient recruitment and retention for PD trials through remote or virtual clinical trials. Virtual trials incorporating telemedicine and mobile applications have been found conducive to improving remote participation and retention in clinical trials for PD.⁶³ Studies like “AT-HOME PD” have successfully recruited diverse participants across geographic locations and received high participant interest in future remote studies.⁶⁴

As more DHTs are adopted, data comparability issues may arise. Data collected from different systems and tools may not follow the same structure, so it may become challenging to integrate and analyze them, causing data comparability issues. Standardized data collection protocols, data storage, and data analysis are crucial to address this. A metadata framework has been recently proposed to standardize the applications of DHTs in PD trials, enhancing data comparability and supporting regulatory maturity of other drug development tools.⁶⁵ While the framework offers valuable guidance, its implementation remains challenging due to the rapid evolution of digital technologies, various trial designs, and diverse patient populations.

Conclusions

Digital technologies and AI have shown promise in improving the success of clinical drug trials by having more accurate targets, more suitable candidates, more efficiency in pre-clinical evaluation, and greater drug monitoring.

The current implementation of DHTs and AI-based tools in clinical drug trials is still in the early stages, but early promising results suggest that their integration could lead to great efficiency of drug development for PD, thereby potentially higher chance of success in clinical drug trials. More exploration is needed regarding the underlying mechanisms of neurodegeneration in PD, the use of existing tools in facilitating drug development at later stages (such as Phase II-III clinical trials, sales and post-market monitoring) and the development of AI-powered new models that are tailored for the development of drugs for different levels of PD and different population groups, with high-quality, real-life data. This paper presented an overview of existing challenges to clinical drug trials, recent digital approaches to improve trial success, and remaining challenges to be resolved in future. Due to its limited length, this review did not intend to synthesize all studies related to AI-powered drug discovery and development for PD. A future systematic review is encouraged to compare the performance of relevant AI models, types of data used by the models and their use cases to further understanding of failures in drug trials for PD. Furthermore, the successful implementation of AI in clinical drug trials requires close collaboration between various stakeholders, including clinicians, patients, data scientists, and regulatory authorities. The process of actual collaboration amongst stakeholders and related challenges also requires future exploration. Suggestions in this work have been made to inform studies in this area to continue the success in the development of symptomatic drugs and further the exploration and discovery of disease-modifying treatment for PD, with lessons learned for future success.

Footnotes

Acknowledgments

The authors have no acknowledgments to report.

ORCID iDs

Cen Cong

Madison Milne-Ives

Ananya Ananthakrishnan

Walter Maetzler

Edward Meinert

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: CC, EM, MM and AA are supported by the National Institute for Health and Care Research (NIHR) Newcastle Biomedical Research Centre (BRC) based at the Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle University, and the Cumbria, Northumberland, and Tyne and Wear (CNTW) NHS Foundation Trust. The views expressed in this publication are those of the author(s) and not necessarily those of NIHR Newcastle BRC, Newcastle Hospitals NHS Trust, or any of the authors' affiliated universities. The funding body was not involved in the study design, data collection or analysis, or the writing and decision to submit the article for publication.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Kolluri

Lin

Liu

, et al. Machine learning and artificial intelligence in pharmaceutical research and development: a review. AAPS J 2022; 24: 19.

Kasoju

Remya

Sasi

, et al. Digital health: trends, opportunities and challenges in medical devices, pharma and bio-technology. CSI Trans ICT 2023; 11: 11.

Office of the Commissioner. Focus Area: Digital Health Technologies. U.S. Food and Drug Administration, https://www.fda.gov/science-research/focus-areas-regulatory-science-report/focus-area-digital-health-technologies (2024, accessed 25 November 2024).

Czech

Badley

Yang

, et al. Improved measurement of disease progression in people living with early Parkinson's disease using digital health technologies. Commun Med 2024; 4: 49.

Bertha

Alaj

Bousnina

, et al. Incorporating digitally derived endpoints within clinical development programs by leveraging prior work. NPJ Digit Med 2023; 6: 139.

Ethics and governance of artificial intelligence for health: WHO guidance Executive summary, https://www.who.int/publications/i/item/9789240037403 (2021, accessed 27 January 2025).

Khanna

Adams

Antoniades

, et al. Accelerating Parkinson’s disease drug development with federated learning approaches. NPJ Parkinsons Dis 2024; 10: 225.

Rehman

, et al. Role of artificial intelligence in revolutionizing drug discovery. Fundamental Research 2024. [Pre-proof]. DOI: https://doi.org/10.1016/j.fmre.2024.04.021.

Umscheid

Margolis

Grossman

. Key concepts of clinical trials: a narrative review. Postgrad Med 2011; 123: 194–204.

10.

Sun

Gao

, et al.

Why 90% of clinical drug development fails and how to improve it?

Acta Pharm Sin B 2022; 12: 3049–3062.

11.

Zahoor

Shafi

Haq

. Pharmacological treatment of Parkinson's disease. In: Stoker

Greenland

(eds) Parkinson’s disease: Pathogenesis and clinical aspects. Brisbane (AU): Codon Publications, 2018.

12.

Gray

Patel

Ives

, et al. Long-term effectiveness of adjuvant treatment with catechol-O-methyltransferase or monoamine oxidase B inhibitors compared with dopamine agonists among patients with Parkinson disease uncontrolled by levodopa therapy: the PD MED randomized clinical trial. JAMA Neurol 2022; 79: 131–140.

13.

Visser

de Vries

Richard

, et al. Tackling vascular risk factors as a possible disease modifying intervention in Parkinson’s disease. NPJ Parkinsons Dis 2024; 10: 50.

14.

Zhao

Cheng

Zhong

. Implications of successful symptomatic treatment in Parkinson’s disease for therapeutic strategies of Alzheimer's disease. ACS Chem Neurosci 2019; 10: 922–930.

15.

Mari

Mestre

. The disease modification conundrum in Parkinson's disease: failures and hopes. Front Aging Neurosci 2022; 14: 810860.

16.

Wolff

Schumacher

Pürner

, et al.

Parkinson's disease therapy: what lies ahead?

J Neural Transm 2023; 130: 793–820.

17.

Lee

Jung

Park

, et al. Biomimetically engineered amyloid-shelled gold nanocomplexes for discovering α-synuclein oligomer-degrading drugs. ACS Appl Mater Interfaces 2023; 15: 2538–2551.

18.

Chen

Menezes

Rodgers

, et al. Small molecule inhibitors of α-synuclein oligomers identified by targeting early dopamine-mediated motor impairment in C. elegans. Mol Neurodegener 2021; 16: 77.

19.

Berven

Kverneng

Sheard

, et al. NR-SAFE: a randomized, double-blind safety trial of high dose nicotinamide riboside in Parkinson's disease. Nat Commun 2023; 14: 7793.

20.

Schiff

Migliori

Chen

, et al. Integrating deep learning and unbiased automated high-content screening to identify complex disease signatures in human fibroblasts. Nat Commun 2022; 13: 1590.

21.

Vaswani

Tropea

Dahodwala

. Overcoming barriers to Parkinson disease trial participation: increasing diversity and novel designs for recruitment and retention. Neurotherapeutics 2020; 17: 1724–1735.

22.

Mathur

DeWitte

Robledo

, et al. Rising to the challenges of clinical trial improvement in Parkinson's disease. J Parkinsons Dis 2015; 5: 263–268.

23.

Vatansever

Schlessinger

Wacker

, et al. Artificial intelligence and machine learning-aided drug discovery in central nervous system diseases: state-of-the-arts and future directions. Med Res Rev 2021; 41: 1427–1473.

24.

Pushpakom

Iorio

Eyers

, et al. Drug repurposing: progress, challenges and recommendations. Nat Rev Drug Discov 2019; 18: 41–58.

25.

Gupta

Kumari

Senapati

, et al. New era of artificial intelligence and machine learning-based detection, diagnosis, and therapeutics in Parkinson’s disease. Ageing Res Rev 2023; 90: 102013.

26.

Lin

Zhang

, et al. Multi-dimensional deep learning drives efficient discovery of novel neuroprotective peptides from walnut protein isolates. Food Funct 2023; 14: 6969–6984.

27.

Di Stefano

Galati

Ortore

, et al. Machine learning-based virtual screening for the identification of Cdk5 inhibitors. Int J Mol Sci 2022; 23: 10653.

28.

Liu

Peng

, et al. Identification of potential Parkinson's disease drugs based on multi-source data fusion and convolutional neural network. Molecules 2022; 27: 4780.

29.

Shin

Lim

, et al. Human motor system-based biohybrid robot-on-a-chip for drug evaluation of neurodegenerative disease. Adv Sci 2024; 11: e2305371.

30.

Christie

Abbas

Koesmahargyo

. Technology for measuring and monitoring treatment compliance remotely. J Parkinsons Dis 2021; 11: S77–S81.

31.

Moreau

Rouaud

Grabli

, et al. Overview on wearable sensors for the management of Parkinson's disease. NPJ Parkinsons Dis 2023; 9: 153.

32.

D'mello

Shivasharanappa

Hanchinalmath

, et al. Chapter 10 - in silico approaches in drug discovery for SARS-CoV-2. In: Egbuna

(ed.) Coronavirus drug discovery (Volume 3: Druggable Targets and In Silico Update). Amsterdam: Elsevier, 2022; 235–251.

33.

Xin

Niu

Tian

, et al. Identification of potential immune-related hub genes in Parkinson’s disease based on machine learning and development and validation of a diagnostic classification model. PLoS One 2023; 18: e0294984.

34.

Horne

Andrzejewska

Alam

, et al. Discovery of potent inhibitors of α-synuclein aggregation using structure-based iterative learning. Nat Chem Biol 2024; 20: 634–645.

35.

Cheng

Wang

Tang

, et al. Artificial intelligence and open science in discovery of disease-modifying medicines for Alzheimer's disease. Cell Rep Med 2024; 5: 101379.

36.

Pedersen

Lange

Førland

MGG

, et al. A systematic review of associations between common SNCA variants and clinical heterogeneity in Parkinson’s disease. NPJ Parkinsons Dis 2021; 7: 54.

37.

Jensen

Schlossmacher

Stefanis

. Who ever said it would be easy? Reflecting on two clinical trials targeting α-synuclein. Mov Disord 2023; 38: 378–384.

38.

Murugan

Muvva

Jeyarajpandian

, et al. Performance of force-field- and machine learning-based scoring functions in ranking MAO-B protein-inhibitor complexes in relevance to developing Parkinson’s therapeutics. Int J Mol Sci 2020; 21: 7648.

39.

Ivanisevic

Sewduth

. Multi-omics integration for the design of novel therapies and the identification of novel biomarkers. Proteomes 2023; 11: 34.

40.

Qiao

Wang

. Causal enhanced drug-target interaction prediction based on graph generation and multi-source information fusion. Bioinformatics 2024; 40: btae570.

41.

Sahota

. Causal AI: Bridging the gap between correlation and causation. Neil Sahota, https://www.neilsahota.com/causal-ai-bridging-the-gap-between-correlation-and-causation/ (2024, accessed 31 January 2025).

42.

Laifenfeld

Yanover

Ozery-Flato

, et al. Emulated clinical trials from longitudinal real-world data efficiently identify candidates for neurological disease modification: examples from Parkinson’s disease. Front Pharmacol 2021; 12: 631584.

43.

Latif

Ullah

Shkodina

, et al. Drug reprofiling history and potential therapies against Parkinson’s disease. Front Pharmacol 2022; 13: 1028356.

44.

Tayebi

BabaAli

. EKGDR: an end-to-end knowledge graph-based method for computational drug repurposing. J Chem Inf Model 2024; 64: 1868–1881.

45.

Romano

Truong

Kumar

, et al. The Alzheimer's knowledge base: a knowledge graph for Alzheimer disease research. J Med Internet Res 2024; 26: e46777.

46.

Tandra

Yoone

Mathew

, et al. Literature-based discovery predicts antihistamines are a promising repurposed adjuvant therapy for Parkinson’s disease. Int J Mol Sci 2023; 24: 12339.

47.

Johnston

Lacoste

AMB

Ravenscroft

, et al. Using artificial intelligence to identify drugs for repurposing to treat l-DOPA-induced dyskinesia. Neuropharmacology 2024; 248: 109880.

48.

Kim

J-B

Kim

S-J

, et al. Artificial intelligence-driven drug repositioning uncovers efavirenz as a modulator of α-synuclein propagation: implications in Parkinson's disease. Biomed Pharmacother 2024; 174: 116442.

49.

Mukherjee

Roy

Ghosh

, et al. Role of animal models in biomedical research: a review. Lab Anim Res 2022; 38: 18.

50.

Athauda

Foltynie

. Drug repurposing in Parkinson's disease. CNS Drugs 2018; 32: 747–761.

51.

Shiammala

Duraimutharasan

NKB

Vaseeharan

, et al. Exploring the artificial intelligence and machine learning models in the context of drug design difficulties and future potential for the pharmaceutical sectors. Methods 2023; 219: 82–94.

52.

Subasri

Krishnan

Dhalla

, et al. Diagnosing and remediating harmful data shifts for the responsible deployment of clinical AI models. bioRxiv 2023; 2023.03.26.23286718 [Preprint]. Posted March 29, 2023.

53.

Javed

El-Sappagh

Abuhmed

. Robustness in deep learning models for medical diagnostics: security and adversarial challenges towards robust AI applications. Artif Intell Rev 2024; 58: 1–107.

54.

Giuffrè

Shung

. Harnessing the power of synthetic data in healthcare: innovation, application, and privacy. NPJ Digit Med 2023; 6: 186.

55.

Waisberg

Ong

Kamran

, et al. Transfer learning as an AI-based solution to address limited datasets in space medicine. Life Sci Space Res (Amst) 2023; 36: 36–38.

56.

Subbaswamy

Chen

Saria

. A unifying causal framework for analyzing dataset shift-stable learning algorithms. J Causal Inference 2022; 10: 64–89.

57.

Polo

Izbicki

Lacerda

Jr , et al. A unified framework for dataset shift diagnostics. arXiv [stat.ML], http://arxiv.org/abs/2205.08340 (2022).

58.

Chang

Y-H

Periñan

Wilson

, et al. AccessPD as a next generation registry to accelerate Parkinson's disease research. NPJ Parkinsons Dis 2024; 10: 66.

59.

Stephenson

Alexander

Aggarwal

, et al. Precompetitive consensus building to facilitate the use of digital health technologies to support Parkinson disease drug development through regulatory science. Digit Biomark 2020; 4: 28–49.

60.

Mittermaier

Venkatesh

Kvedar

. Digital health technology in clinical trials. NPJ Digit Med 2023; 6: 88.

61.

PKG Health. PKG Health, https://pkghealth.com/ (2023, accessed 31 January 2025).

62.

Lipsmeier

Taylor

Postuma

, et al. Reliability and validity of the roche PD Mobile application for remote monitoring of early Parkinson’s disease. Sci Rep 2022; 12: 12081.

63.

Salve

Mude

Tarale

, et al. Telemedicine and remote clinical trials: revolutionizing healthcare research. In: 2024 International Conference on Healthcare Innovations, Software and Engineering Technologies (HISET), pp.108–111. IEEE.

64.

Myers

Augustine

Baloga

, et al. Recruitment for remote decentralized studies in Parkinson's disease. J Parkinsons Dis 2022; 12: 371–380.

65.

Hill

Stephenson

Brayanov

, et al. Metadata framework to support deployment of digital health technologies in clinical trials in Parkinson's disease. Sensors 2022; 22: 2136.

From past to future: Digital approaches to success of clinical drug trials for Parkinson's disease

Abstract

Plain language summary

Keywords

Background

Digital approaches to drug discovery and development

Novel targets and inhibitors

Drug repurposing and candidate selection

DHTs for greater patient participation and drug monitoring

Conclusions

Footnotes

Acknowledgments

ORCID iDs

Funding

Declaration of conflicting interests

References