Handbook of Digital Technologies in Movement Disorders

By Roongroj Bhidayasiri

Over the past few years, there have been fundamental changes in the diagnosing and treating patients with chronic diseases, significantly affecting management of neurological movement disorders. In addition, the health and fitness sector developed several devices to better classify, track, and potentially treat chronic diseases. Both handling and interpreting these large datasets has been revolutionized, by machine and deep learning approaches, leading to new and more effective therapies, resulting in longer survival rates.

Handbook of Digital Technologies in Movement Disorders aims to unite these factors to provide a comprehensive guide to patient focused treatments for movement disorders. This book is divided into five distinct sections, starting with an introduction to digital technologies, concepts, and terminologies. The following section reviews various perspectives on technology in movement disorders, including patient and medical professionals. The third section presents technologies used in detecting, measuring progression, and determining response to treatments. This is followed by reviewing the technology used in various treatments of movement disorders including assistive and robotic technologies. Finally, the last section examines the challenges with technology including privacy and other ethical issues.

• Reviews different stakeholders' perspectives on technology in movement disorders
• Presents technological advancements for diagnosing, monitoring, and managing Parkinson’s disease
• Discusses challenges with implementing technology into treatment

• Language: English
• Publisher: Academic Press
• Release date: Jan 20, 2024
• ISBN: 9780323994958

Book preview

Part I

Digital technologies: The primer

Outline

Chapter 1. We are living in the Parkinson’s pandemic: how can multi-stakeholders utilize digital technologies effectively?

Chapter 2. Embracing the promise of artificial intelligence to improve patient care in movement disorders

Chapter 1: We are living in the Parkinson's pandemic

how can multi-stakeholders utilize digital technologies effectively?

Roongroj Bhidayasiria,b, Soania Mathurc, and Walter Maetzlerd     aChulalongkorn Centre of Excellence for Parkinson's Disease & Related Disorders, Department of Medicine, Faculty of Medicine, Chulalongkorn University and King Chulalongkorn Memorial Hospital, Thai Red Cross Society, Bangkok, Thailand     bThe Academy of Science, The Royal Society of Thailand, Bangkok, Thailand     cUnshakeable MD, Toronto, ON, Canada     dDepartment of Neurology, Kiel University and University Hospital Schleswig-Holstein, Campus Kiel, Kiel, Germany

Abstract

The burden of Parkinson's disease (PD), as a prototypical example of movement disorders, continues to grow at an unprecedented pace, recently declared by the WHO as the fastest growing neurodegenerative disorder in terms of prevalence, disability, and death. In many ways, this surge in PD exhibits many of the characteristics of a pandemic (widespread, migration, and high burden), and coupled with the recent COVID-19 pandemic, this double crisis has impacted the care of PD and movement disorders in numerous ways, diametrically opposing the direction our healthcare systems have been taking over the past years. The traditional passive approach to just wait for symptomatic patients to seek specialist services that are scant in many parts of the world is not going to be sustainable. The challenges we face are valuable lessons that constantly remind us that we cannot spend each day fixing errors we all made yesterday. We propose, as detailed in this chapter and several parts in this handbook, that digital technologies could be part of the solution. While advances are evident in terms of better understanding of diseases through a surge of scientific publications in the field of digital technologies in movement disorders, clinical implementations are profoundly lacking, indicating a significant gap among the various stakeholders who are involved in the care of patients with movement disorders. Most importantly, patients are not centrally and frequently involved in the roadmap for implementation of patient-centered digital outcome measures. While differences exist in each region, country, society, and system, a patient-centric multidisciplinary approach is needed to analyze and understand the challenges at different levels and utilize digital technologies as a vehicle, or medium, to integrate fragmented care, and enhance care delivery and interdisciplinary collaboration. Embracing digital technologies this way will accelerate opportunities for early (and remote) diagnosis and intervention, including prevention of disease in high-risk individuals, protocol-driven healthcare to improve quality of care, and better equity of access to treatment and services.

Keywords

Digital biomarkers; Digital phenotyping; Digital technology; Parkinson's disease; Parkinson pandemic; Patient-centered digital outcome measures; Remote diagnosis

Parkinson's disease as a prototypical hypokinetic disorder: recognized for more than 200 years we now know more and more about it

Living in the era of a double (Parkinson and COVID-19) pandemic: an upcoming global challenge

There is no one size-fits-all solution with differences in each region, country, society, and system: where do digital technologies stand?

Digital solutions in Parkinson's disease and other movement disorders: global, community, and individual considerations

There are things that we know about PD and movement disorders, and there are also many unknowns, but we should balance the use of digital technologies to find the answers to what we know we want to know and what matters to patients

Toward a real-life implementation of clinically relevant multi-modal digital markers

Conclusion

References

List of abbreviations

ADL   Activity of Daily Living

COVID-19   Coronavirus Disease of 2019

HIC   High-Income Countries

LIC   Low-Income Countries

PCDO   Patient-Centered Digital Outcome

PD   Parkinson's Disease

QoL   Quality of Life

US   United States of America

WHO   World Health Organization

Parkinson's disease as a prototypical hypokinetic disorder: recognized for more than 200 years we now know more and more about it

With 100 billion neurons and 100 trillion neuronal connections, the brain is justifiably considered the most complicated thing in the universe and when there is a disease process that begins to affect these complex connections, finding a way to treat or even identify the resulting pathology can become exceedingly difficult. Examples include movement disorders where dysfunction of these complex connections (circuitries) can result in different forms of abnormal movements, primarily classified as hypokinetic (i.e., parkinsonism) and hyperkinetic disorders (i.e., tremor, dystonia, chorea), and sometimes in a combination (Fahn, 2011). Parkinson's disease (PD) is a prototypical hypokinetic disorder where degeneration gradually affects different circuitries within the central nervous system that regulate various neurotransmitters, not limited to dopamine, resulting in a slowly progressive natural history, beginning with a long prodromal period, followed by cardinal motor features (slow movement, tremor, rigidity, walking difficulty and imbalance), associated with a wide variety of nonmotor symptoms (neuropsychiatric and autonomic symptoms, disorders of sleep and wakefulness, and pain) (Parkinson disease, 2022). As originally described by James Parkinson in his essay on Shaking Palsy in 1817 (Parkinson, 1817), most clinical features he described still hold true to this day, indicating that PD is probably unchanged, or without significant change, as a disease that affects individual patients. What has changed over the past 200 years is our understanding of the disease's pathophysiological mechanism, etiology, diagnostic methods, clinicopathological correlations, and therapeutic options, which, in combination, impact how the disease, once initially recognized as a rare disorder, recorded as causing only 22 deaths out of 15 million people in England and Wales in 1855, has evolved into the disorder with the fastest growing in age-standardized prevalence, disability, and deaths among all neurological disorders, with over six million patients globally (Dorsey et al., 2018; The Global Burden of Disease, Injuries, and Risk Factors Study GBD 2016 Parkinson’s Disease Collaborators, 2018). It is projected that the number of PD patients will double from 4.6 million in 2005 to 9.3 million by 2030, and continue to rise to over 12 million by 2040 (Disease et al., 2018; Dorsey et al., 2007).

Living in the era of a double (Parkinson and COVID-19) pandemic: an upcoming global challenge

There are probably several, both known and unknown, factors that influence the unprecedented pace of the increase in PD population. The first undeniable factor is global aging, defined as those aged 60 years or over, currently amounting to 670 million in 2022, which represents 14% of the total population with a projection to increase to 26% by 2050 (ESCAP, 2022). As the process of senescence is strongly correlated with PD pathology, the rise in the PD population is primarily influenced by the effects of aging with additional contributions from increasing longevity, declining smoking rates, and impacts of industrialization. In principle, the number of people with a disease is a function of the incidence of the disease and the survival of those with the condition. Therefore, an increase in global PD population is affected by its incidence, which is found to be low before the age of 50 years, but increases rapidly with age, peaking in most studies at around 80 years (Ascherio & Schwarzschild, 2016). As people live longer with a mean life expectancy of 73.2 years in 2022, markedly increased from 47 years in 1950, we are, as a consequence, seeing a larger aging population with PD (Life expectancy of the world, 2022). In addition, these days PD patients also live longer, as evident by a 12% increase in the age-standardization prevalence during the past 20 years which is concurrent with levodopa becoming widely available (DALYs et al., 2015; Hobson et al., 2010; Wanneveich et al., 2018). However, the reduction in mortality is possibly not only due to the availability of pharmacological management, but also the introduction of specialist management services, multidisciplinary team interventions and early identification of PD symptoms. Increased public awareness of PD and availability of telehealth can also lead to earlier in-person and/or virtual specialist visits, potentially contributing to earlier and more accurate diagnosis, especially in the areas where specialist services are underrepresented (Bhidayasiri, Virameteekul, Sukoandari, et al., 2020; Fall et al., 2022; Safiri et al., 2023). Overall, it seems that PD fulfills the criteria for a disease pandemic; it's on a rise in every major region of the world with a disease burden shifting from the West to the East (Dorsey et al., 2018). The only difference that PD has from other pandemic diseases is that PD is a noninfectious disorder so no one can develop immunity to this condition.

As the burden of PD and other movement disorder conditions continues to increase, our healthcare sectors are facing an ever increasing demand for specialist assessments, advanced treatments, and rehabilitation as well as other supporting services. However, this need for medical care is often unmatched, restricted by very limited resources, mainly inaccessibility to treatments and a shortage of qualified healthcare providers, particularly neurologists. Indeed, the lack of trained personnel is recognized as a worldwide problem. According to the WHO Neurology Atlas in 2017, there were only 0.03 neurologists per 100,000 in low-income countries (LIC) and, while high-income countries (HICs) had a higher number of neurologist (4.74 neurologists:100,000 population) than LIC, only a minority (23%) reside and work in rural areas (Atlas: Country resources for neurological, 2017). Even in the US, where specialist training is undertaken, the USA's National Center for Health Workforce Analysis estimated that the expected supply of US neurologists predicted to grow by 11% between 2013 and 2025, will not meet the expected 16% increase in demand for their services (Burton, 2018; Dall et al., 2013). In terms of medication availability, levodopa, which remains the most effective medication for PD, was consistently available in primary care in only 34% of the 100 countries surveyed, none of which were LICs. Limited access to levodopa means that people with PD cannot be given even the most basic therapy to improve their quality of life. This is not to mention the other essentials for optimal management of PD, which include interdisciplinary care teams, specialist treatment centers, rehabilitation and palliative services. Then in 2019, on top of this mismatch in resources, the Coronavirus disease 2019 (COVID-19) hit all of us by surprise, resulting in a major pandemic, impacting the care of PD and movement disorders in numerous ways with radical changes imposed on the structure of healthcare and support systems globally (Bhidayasiri, Virameteekul, Kim, et al., 2020; Poonja et al., 2022). Across the world, it became clear our healthcare systems were not designed to deal with such a crisis, especially unpredictable, large-scale health challenges that require urgent mobilization of resources and affect the whole PD population. In several regions, fragile healthcare systems were stressed to the brink of breakdown or collapse, potentially delaying proper management of PD and movement disorder patients, even those in acute or emergency situations (Grimaldi et al., 2022; Schirinzi et al., 2020).

There is no one size-fits-all solution with differences in each region, country, society, and system: where do digital technologies stand?

This chapter starts with global statistics for PD as a representative of movement disorder conditions to highlight the current situation and the need for urgent public health action at a global level to meet the health and social requirements of people with movement disorders (Schiess et al., 2022). The impact of the double pandemic alerts us to learn from the mistakes of the past and rethink our current healthcare systems. How can an already resource-constrained healthcare system (re)configure to improve its ability to handle large-scale health issues on movement disorders while remaining sustainable? The answer is not to address individual challenges (i.e., training more neurologists and healthcare professionals to treat more PD patients with advanced conditions) without first looking at the global picture, bearing in mind that a one size-fits-all solution is unlikely to be found—each region, country, society, and system will need to adapt differently to fix the unmet needs. It is imperative to think about these fundamental questions around the current situation and challenges in PD and other movement disorders to inform near- and long-term action plans. Fundamentally, PD is a progressive disease with people living with the condition frequently requiring long-term care. An interdisciplinary approach ensures optimal disease management (Bloem et al., 2020). However, providing holistic care may be daunting in low-resource settings, where limited resources impede the building of an integrated system of care along a continuum of services that includes diagnosis, treatment, rehabilitation, and palliative care (Schiess et al., 2022). It is essential that changes are made as the COVID-19 pandemic might just be a rehearsal with more challenges approaching our doors, such as a link between climate change and health issues, including PD (Bongioanni et al., 2021).

Digital solutions in Parkinson's disease and other movement disorders: global, community, and individual considerations

The authors in this handbook do not aim to provide a single comprehensive answer to address these global challenges, but rather provide our perspective on how we, as a group of multiple stakeholders (neurologists, patients, researchers, engineers and developers, policy makers) involved in the care of PD and other movement disorders, can accelerate digitization into interdisciplinary management to form part of the solution to the various difficulties we are facing similarly or differently in each society. For example, LICs may want to focus on strengthening health and social systems and building capacity to ensure the availability of essential drugs, diagnostics and interdisciplinary therapies in movement disorders, whereas HICs, where treatment availability is not a predominant issue, may want to focus their resources in developing effective strategies for disease prevention and risk reduction (Schiess et al., 2022). Nevertheless, these strategies require collaborative understanding of various stakeholders to analyze, understand, and harmonize these approaches into an effective global action to improve the care and management of PD and movement disorders. At a global level, where challenges in PD management were first highlighted, the WHO identified six workable avenues for action: (1) disease burden estimation; (2) advocacy and awareness; (3) prevention and risk reduction; (4) diagnosis, treatment, and care; (5) caregiver support; and (6) research to address global disparities in PD, with digital technologies able to be incorporated into action plans for most of these avenues (Schiess et al., 2022). Tremendous potential has been demonstrated in the fields of PD screening and subtype classification, where current technologies are capable of differentiating affected individuals from healthy controls and clustering patients based on disease severity in a large population with an acceptable accuracy (Deb et al., 2022; Rana et al., 2022; Surangsrirat et al., 2022). At a community level, although we have gained substantial insight into the pathophysiology and also partly into the etiology of PD and other movement disorders, current outcome measures for diagnosis, and assessment of progression and treatment response are a variable mix of clinical judgment, scaled rating scores and retrospective patient reporting. Both patients and treating clinical teams recognize that this way of evaluating the disease is not sufficient. For patients, although the greatest desire is for a cure for their disease, until that disease modifying treatment is found, optimizing quality of life (QoL) is a key focus with the understanding that movement disorders can affect a patient on many levels—physically, emotionally, mentally, and socially. The degree to which each domain is affected and what constitutes a good QoL differs for every patient, depending on factors such as age, disease stage, and circumstances of life. For example, it has been shown that the concerns of young-onset PD can be quite different from those with late-onset PD due to their unique clinical symptoms (i.e., dystonia, levodopa-induced dyskinesias, anxiety and depression) and different family and social engagements, which requires a tailored multidisciplinary approach (Post et al., 2020). While we are aiming for early diagnosis of movement disorders, an early and timely diagnosis may be preferable to the individual patient, especially where the healthcare system is able to provide support and management that is available for the individuals at every stage of the disease (Rees et al., 2018). Another issue for consideration is the variability that also exists in the same patient at different points in their journey with PD. Therefore, outcome measures must focus on what is meaningful for an individual patient's life experience. Important measures such as monitoring patients in their home, work, or social environment, can provide insights into the patient's lived experience, and are much more relevant than the snapshots of their condition that is garnered at intermittent clinic visits (Maetzler et al., 2021). With this in mind, our ongoing search for a better understanding of PD and other movement disorders has spurred new approaches and innovations, taking advantage of the advancements in technology over the last 2 decades (Deb et al., 2022). Indeed, great effort is being put into research within our clinical communities to harness the power of digital technologies to assist the drive toward optimal clinical management and better treatments of these disorders. Recent work has also been put into the development of digital outcomes that capture small changes over at least 1 year in early PD patients or those in a prodromal stage, which is the duration of most disease-modifying trials (Mirelman et al., 2022). Various digital technologies are now being used or tested to assist with diagnosis, improve measurement of symptoms and to support disease management with a roadmap proposed for their implementation using mobile health technologies (Espay et al., 2019). For example, digital technology can be used to measure motor features and predict clinical events (Adams et al., 2021), and provide long-term monitoring and telemedical consulting using automated devices, which has already enable a better understanding of movement disorders (Salchow-Hommen et al., 2022). In addition, digital assistive technology for the (nonmedical) treatment of symptoms in movement disorders is increasingly finding its way into clinical practice (Bhidayasiri, 2020; Bhidayasiri et al., 2022), and there is increasing evidence that nonmotor symptoms—which are also common in movement disorders—can be effectively measured using digital technology (van Wamelen et al., 2021).

There are things that we know about PD and movement disorders, and there are also many unknowns, but we should balance the use of digital technologies to find the answers to what we know we want to know and what matters to patients

It is highly likely that these results and successes with digital technology use in movement disorders are only the beginning of a fundamental development that could ultimately lead to a paradigm shift in how we diagnose, evaluate and treat these diseases. However, it is important to remember, as put succinctly by William Bruce Cameron and later by Albert Einstein, that Not everything that can be counted counts and not everything that counts can be counted (Mathur, 2021). We should not be using these new technologies to measure and monitor anything and everything, but rather we should monitor what matters to patients. We have already recognized that with such digital tools we can objectively capture a new dimension of human life that is affected by the disease, that of daily life, real-life. Effectively evaluating this elusive dimension has previously evaded researchers, despite being a fundamental marker of a disease's impact on a person's well-being. By using digital technology tailored to objectively monitor changes in these areas, new therapies and treatment can be developed that improve QoL.

If we are going to capture the real life impact of new digital technologies in PD and other movement disorders, it is advisable that we involve patients from the beginning and centrally in the development. As the manifestations of PD is so individualized, determining clinically actionable phenotypes for digital phenotyping that lead to patient-centered digital outcome measures (PCDOs) in an individual patient should be based on a joint decision by the team that is involved in the care of that patient. This approach provides a transition from focusing on single symptom domains (e.g., tremor or freezing of gait) to patient-centered management of how the complex symptomatologies of individual patient impair his/her activities of daily living (ADLs) and QoL. Recently, a framework has been developed to define the scope of PCDOs that refers to technology-based outcomes that facilitate clinically important decision-making by the clinician and promote long-term adherence by the patient (Espay et al., 2019). These digital outcomes should be sensitive to individual patient preferences, needs and values, ensuring that patient values guide all clinical decisions. This initiative is very promising and new developments are being closely followed by our patients.

Toward a real-life implementation of clinically relevant multi-modal digital markers

Up until now, ideal disease management is based on the identification of symptoms and subsequent treatment protocols focus on the optimal control of those symptoms in a way that is meaningful for the patient's life experience. In diseases like hypertension for example, there is a known optimal range of blood pressure that clinicians aim to reach, and researchers use as their goal, as those values have been shown to reduce risk of cardiovascular complications and improve survival. Unfortunately, in PD and other movement disorders, we have yet to agree upon accepted markers to use as targets for treatment. The authors here argue that these targets should ideally be single- or multi-component markers collected in the home environment. Only then can they adequately reflect the impact of the disease on a patient's real life—the actual QoL. Thanks to advances in digital technology, this ideal is partially possible, providing real-time monitoring and information gathering about movement disorder-relevant aspects, such as motor symptoms and mobility limitations, wherever a patient is, removing the need for complicated and clinical environments. This is the concept of digital phenotyping, defined as a moment-by-moment quantification of the individual-level patient phenotype in situ using data from personal digital devices, most commonly with smartphones (Bhidayasiri & Mari, 2020). Originally adopted in behavioral studies, the concept of digital phenotypes has been increasingly used in PD and other movement disorders, such as Huntington's disease (Torous & Keshavan, 2018). However, it should be kept in mind that currently the assessment of other aspects of daily life, including mental health and socialization, is still quite difficult even with the most advanced digital technologies.

Another requirement that we have always had for the development of markers for Parkinson's and other movement disorders, but which could be comprehensively addressed for the first time by digital technology, is the many faces that these diseases can have, both on inter and intraindividual levels. The degree to which different human domains are affected and what constitutes quality of life differs from patient to patient, depending on factors such as age and circumstances of life. This variability also exists in the same patient at different disease phases. The best outcome measure for clinical care and for clinical trials may therefore be one that is personalized and tailored to each individual patient based on their QoL goals. For some, it may be continuing to run marathons, for others it may be navigating around their home without falling. It is therefore of central importance to consider, in the context of the development of digital markers for the measurement of disease progression and treatment response in PD and other movement disorders, whether an individualization of markers can be accepted, as is already the case in other diseases such as cerebral palsy (both for clinical care and for trials) (Verkerk et al., 2006; Wessels et al., 2001).

It is conceivable, and also hoped, that patients will use the development in this digital area to increase their engagement and involvement in interactions with their medical professional team. Patients can use new digital technologies to engage in self-monitoring and self-care, and they can, and should, be involved as a central part of networks involved in the development of these new digital markers and therapies (Lupton, 2013; Milne-Ives et al., 2022). Seeking such engagement not only enable us to motivate people living with PD and other movement disorders to become actively involved in research, but also has the knock-on effect of improving the research outcomes (Port et al., 2021). By being involved in their own care, patients are uniquely positioned to identify areas of concern and help choose most relevant outcome measures. Just as there should be shared decision making between patients and all other necessary stakeholders in personal disease management, there should also be collaboration in clinical observational studies and trials from inception to design, recruitment and communication of the results (Stephenson et al., 2021).

Conclusion

For all the reasons mentioned above, the topic of digital technologies in movement disorders has attracted a growing interest over the last 2 decades from patients, their treating teams, developers of new outcome parameters and therapies in academia and industry, regulatory authorities and policy makers. This is reflected in the virtually exponential growth of scientific literature for this topic area over this time span (Deb et al., 2022). It is therefore difficult to keep a good overview of the developments in this multi-stakeholder adventure. This book aims to address this: it allows the different stakeholders involved to express their view on the current state-of-the-art of digital technologies in movement disorders, and to shed light on the future of this topic from their respective sides. The authors have deliberately focused on the everyday relevance of development of digital technologies in movement disorders, tried to integrate the patient perspective wherever possible, and emphasized the patient–professional treating team partnership, as it is this alliance that will be central to working out and posing the right questions to researchers. By keeping patient voices front and central, developers can go about their development work with confidence and expectation that the end result of a clinically and, most importantly, patient-relevant application will be reached.

References

1. Adams J.L, Lizarraga K.J, Waddell E.M, Myers T.L, Jensen-Roberts S, Modica J.S, Scheider R.B.Digital technology in movement disorders: Updates, applications, and challenges. Current Neurology and Neuroscience Reports. 2021;21(4):16.

2. Ascherio A, Schwarzschild M.A. The epidemiology of Parkinson's disease: Risk factors and prevention. The Lancet Neurology. 2016;15(12):1257–1272.

3. Atlas: Country resources for neurological disorders. 2nd ed. Geneva: World Health Organization; 2017.

4. Bhidayasiri R. Assistive technologies in Parkinson's disease. In: Martin C.R, Preedy V.R, eds. The neuroscience of Parkinson's disease. 1. London: Academic Press; 2020:713–728.

5. Bhidayasiri R, Mari Z. Digital phenotyping in Parkinson's disease: Empowering neurologists for measurement-based care. Parkinsonism and Related Disorders. 2020;80:35–40.

6. Bhidayasiri R, Maytharakcheep S, Phumphid S, Maetzler W. Improving functional disability in patients with tremor: A clinical perspective of the efficacies, considerations, and challenges of assistive technology. Journal of the Neurological Sciences. 2022;435:120197.

7. Bhidayasiri R, Virameteekul S, Kim J.M, Pal P.K, Chung S.J. COVID-19: An early review of its global impact and considerations for Parkinson's disease patient care. Journal of Movement Disorders. 2020;13(2):105–114.

8. Bhidayasiri R, Virameteekul S, Sukoandari B, Tran T.N, Lim T.T. Challenges of Parkinson's disease care in Southeast Asia. In: Riederer P, Laux G, Nagatsu T, Le W, Riederer C, eds. NeuroPsychopharmacotherapy. Cham: Springer International Publishing; 2020:1–21.

9. Bloem B.R, Henderson E.J, Dorsey E.R, Okun M.S, Okubadejo N, Chan P, Andrejack J, Darweesh S.K.L, Munneke M.Integrated and patient-centred management of Parkinson’s disease: A network model for reshaping chronic neurological care. The Lancet Neurology. 2020;19(7):623–634.

10. Bongioanni P, Del Carratore R, Corbianco S, Diana A, Cavallini G, Masciandaro S.M, Dini M, Buizza R.Climate change and neurodegenerative diseases. Environmental Research. 2021;201:111511.

11. Burton A. How do we fix the shortage of neurologists?The Lancet Neurology. 2018;17(6):502–503.

13. Dall T.M, Storm M.V, Chakrabarti R, Drogan O, Keran C.M, Donofrio P.D, Henderson V.H, Kaminski H.J, Stevens J.C, Vidic T.R.Supply and demand analysis of the current and future US neurology workforce. Neurology. 2013;81(5):470–478.

14. DALYs G.B.D, Collaborators H, Murray C.J, Barber R.M, Foreman K.J, Abbasoglu Ozgoren A, Abd-Allah F, Abera S.F, Aboyans V, Abraham J.P, Abubakar I, Abu-Raddad L.J, Abu-Rmeileh N.M.Global, regional, and national disability-adjusted life years (DALYs) for 306 diseases and injuries and healthy life expectancy (HALE) for 188 countries, 1990-2013: Quantifying the epidemiological transition. Lancet. 2015;386(10009):2145–2191.

15. Deb R, An S, Bhat G, Shill H, Ogras U.Y. A systematic survey of research trends in technology usage for Parkinson's disease. Sensors (Basel). 2022;22(15).

16. Disease G.B.D, Incidence I, Prevalence C. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990–2017: A systematic analysis for the global burden of disease study 2017. Lancet. 2018;392(10159):1789–1858.

17. Dorsey E.R, Constantinescu R, Thompson J.P, Biglan K.M, Holloway R.G, Kieburtz K, Marshall F.J, Ravina B.M, Schifitto G, Siderowf A, Tanner C.M.Projected number of people with Parkinson disease in the most populous nations, 2005 through 2030. Neurology. 2007;68(5):384–386.

18. Dorsey E.R, Sherer T, Okun M.S, Bloem B.R. The emerging evidence of the Parkinson pandemic. Journal of Parkinson's Disease. 2018;8(s1):S3–S8.

19. Espay A.J, Hausdorff J.M, Sanchez-Ferro A, Klucken J, Merola A, Bonato P, Paul S.S, Horak F.B, Vizcarra J.A, Mestre T.A, Reilmann R.A roadmap for implementation of patient-centered digital outcome measures in Parkinson's disease obtained using mobile health technologies. Movement Disorders. 2019;34(5):657–663.

20. Fahn S. Classification of movement disorders. Movement Disorders. 2011;26(6):947–957.

21. Fall M, Dardare I.M, Diop A.M, Pelagie M.A, Kahwagi J, Dechacus G.C, Gaye N.M, Rizig M, Diagne N.S, Ndiaye M, Diop A.G.Spectrum of movement disorders: Experience of a one and half year of existence of the first specialized center in Senegal. Parkinsonism and Related Disorders. 2022;98:13–15.

22. Grimaldi S, Eusebio A, Carron R, Regis J.M, Velly L, Azulay J.P, Witjas T. Deep brain stimulation-withdrawal syndrome in Parkinson’s disease: Risk factors and pathophysiological hypotheses of a life-threatening emergency. Neuromodulation. 2022;26(2):424–434.

23. Hobson P, Meara J, Ishihara-Paul L. The estimated life expectancy in a community cohort of Parkinson's disease patients with and without dementia, compared with the UK population. Journal of Neurology Neurosurgery and Psychiatry. 2010;81(10):1093–1098.

24. Life expectancy of the world population. Worldometer; 2022 Available from:. https://www.worldometers.info/demographics/life-expectancy/.

25. Lupton D. The digitally engaged patient: Self-monitoring and self-care in the digital health era. Social Theory and Health. 2013;11(3):256–270.

26. Maetzler W, Rochester L, Bhidayasiri R, Espay A.J, Sanchez-Ferro A, van Uem J.M.T. Modernizing daily function assessment in Parkinson’s disease using capacity, perception, and performance measures. Movement Disorders. 2021;36(1):76–82.

27. Mathur S. Counting what counts: How to reach outcomes that truly matter to Parkinson's patients. Movement Disorders. 2021;36(6):1290–1292.

28. Milne-Ives M, Carroll C, Meinert E. Self-management interventions for people with Parkinson disease: Scoping review. Journal of Medical Internet Research. 2022;24(8):e40181. .

29. Mirelman A, Siderowf A, Chahine L. Outcome assessment in Parkinson disease prevention trials: Utility of clinical and digital measures. Neurology. 2022;99(7 Suppl. 1):52–60.

30. Parkinson J. An essay on the shaking palsy. London: Sheerwood, Neely and Jones; 1817.

31. Parkinson disease: A public health approach. Technical brief. Geneva: World Health Organization; 2022.

32. Poonja S, Chaudhuri K.R, Miyasaki J.M. Movement disorders in COVID-19 times: Impact on care in movement disorders and Parkinson disease. Current Opinion in Neurology. 2022;35(4):494–501.

33. Port R.J, Rumsby M, Brown G, Harrison I.F, Amjad A, Bale C.J. People with Parkinson's disease: What symptoms do they most want to improve and how does this change with disease duration?Journal of Parkinson's Disease. 2021;11(2):715–724.

34. Post B, van den Heuvel L, van Prooije T, van Ruissen X, van de Warrenburg B, Nonnekes J. Young onset Parkinson's disease: A modern and tailored approach. Journal of Parkinson's Disease. 2020;10(s1):S29–S36.

35. Rana A, Dumka A, Singh R, Panda M.K, Priyadarshi N, Twala B. Imperative role of machine learning algorithm for detection of Parkinson's disease: Review, challenges and recommendations. Diagnostics (Basel). 2022;12(8).

36. Rees R.N, Acharya A.P, Schrag A, Noyce A.J. An early diagnosis is not the same as a timely diagnosis of Parkinson's disease. F1000Research. 2018;7.

37. Safiri S, Noori M, Nejadghaderi S.A, Mousavi S.E, Sullman M.J.M, Araj-Khodaei M, Singh K, Kolahi A.A, Gharagozli K.The burden of Parkinson's disease in the Middle East and North Africa region, 1990–2019: Results from the global burden of disease study 2019. BMC Public Health. 2023;23(1):107.

38. Salchow-Hommen C, Skrobot M, Jochner M.C.E, Schauer T, Kuhn A.A, Wenger N. Review-emerging portable technologies for gait analysis in neurological disorders. Frontiers in Human Neuroscience. 2022;16:768575.

39. Schiess N, Cataldi R, Okun M.S, Fothergill-Misbah N, Dorsey E.R, Bloem B.R, Barretto M, Bhidayasiri R, Brown R, Chishimba L, Chowdhary N.Six action steps to address global disparities in Parkinson disease: A World Health Organization priority. JAMA Neurology. 2022;79(9):929–936.

40. Schirinzi T, Cerroni R, Di Lazzaro G, Liguori C, Scalise S, Bovenzi R, Conti M, Garasto E, Mercuri N.B, Pierantozzi M, Pisani A.Self-reported needs of patients with Parkinson's disease during COVID-19 emergency in Italy. Neurological Sciences. 2020;41(6):1373–1375.

41. ESCAP population data sheet 2022. In: Division Sd, ed. United nations economic and social commission for Asia and the Pacific. 2022.

42. Stephenson D, Badawy R, Mathur S, Tome M, Rochester L. Digital progression biomarkers as novel endpoints in clinical trials: A multistakeholder perspective. Journal of Parkinson's Disease. 2021;11(s1):S103–S109.

43. Surangsrirat D, Sri-Iesaranusorn P, Chaiyaroj A, Vateekul P, Bhidayasiri R.Parkinson's disease severity clustering based on tapping activity on mobile device. Scientific Reports. 2022;12(1):3142.

12. The Global Burden of Disease, Injuries, and Risk Factors Study (GBD) 2016 Parkinson’s Disease Collaborators, . Global, regional, and national burden of Parkinson’s disease, 1990–2016: A systematic analysis for the global burden of disease study 2016. The Lancet Neurology. 2018;17(11):939–953.

44. Torous J, Keshavan M. A new window into psychosis: The rise digital phenotyping, smartphone assessment, and mobile monitoring. Schizophrenia Research. 2018;197:67–68.

45. Verkerk G.J, Wolf M.J, Louwers A.M, Meester-Delver A, Nollet F. The reproducibility and validity of the Canadian Occupational Performance Measure in parents of children with disabilities. Clinical Rehabilitation. 2006;20(11):980–988.

46. van Wamelen D.J, Sringean J, Trivedi D, Carroll C.B, Schrag A.E, Odin P, Antonini A, Bloem B.R, Bhidayasiri R, Chaudhuri K.R.Digital health technology for non-motor symptoms in people with Parkinson's disease: Futile or future?Parkinsonism and Related Disorders. 2021;89:186–194.

47. Wanneveich M, Moisan F, Jacqmin-Gadda H, Elbaz A, Joly P. Projections of prevalence, lifetime risk, and life expectancy of Parkinson's disease (2010–2030) in France. Movement Disorders. 2018;33(9):1449–1455.

48. Wessels R, de Witte L, Andrich R, Ferrario M, Persson J, Oberg B, Oortwijn W, VanBeekum T, Lorentsen Ø.IPPA, a user-centred approach to assess effectiveness of assistive technology provision. Technology and Disability. 2001;13(2):105–115.

Chapter 2: Embracing the promise of artificial intelligence to improve patient care in movement disorders

Roongroj Bhidayasiria,b, and Christopher G. Goetzc     aChulalongkorn Centre of Excellence for Parkinson's Disease & Related Disorders, Department of Medicine, Faculty of Medicine, Chulalongkorn University and King Chulalongkorn Memorial Hospital, Thai Red Cross Society, Bangkok, Thailand     bThe Academy of Science, The Royal Society of Thailand, Bangkok, Thailand     cDepartment of Neurological Sciences, Rush University Medical Center, Chicago, IL, United States

Abstract

Artificial Intelligence (AI) is one of the most exciting methodological shifts in our era, offering an enormous potential to transform healthcare with humans and machines working together to provide better patient care. While we continue to see rapid growth of AI applications in healthcare, there is a rising concern that AI will, in the future, outperform or even replace physicians. In our own view, AI and neurologists are most potent when they cooperate, and neurologists must decide how AI will be best integrated into neurological clinical practice. To make it relevant to daily practices, the authors begin this chapter by outlining the benefits of AI in the simulation of clinical reasoning, followed by a four-phase approach to implement AI as an assistant, monitor, coach and teammate with most examples related to Parkinson's disease. Understanding what neurologists working with AI can achieve will ultimately benefit our patients, where impressive gains have been observed in the field of movement disorders, encompassing screening, assessment of severity, and monitoring by enhancing predictive accuracy and augmenting specific diagnostic procedures as long as there is enough data to train the system. However, it seems that AI, in its current configuration, is very good for solving very specific problems, and what AI cannot achieve at this stage is sound clinical judgment, which requires tailoring an approach to what is demanded in a particular clinical circumstance—a capacity for flexible and contextual reasoning (practical wisdom). As a result, neurologists will remain indispensable and should apply quantitative tools, as provided by AI, alongside interpretive methods, exercising practical wisdom to determine which approach is most suited to the particular context where patients' rights are to be respected. Importantly, patients' involvement in decisions about their care is a central premise to the approach that we all should follow.

Keywords

Artificial intelligence; Artificial neural network; Big data; Clinical reasoning; Deep learning; Machine learning; Movement disorders; Neurologists

Introduction

Clinical reasoning in movement disorders: Neurologist's view

Toward the simulation of neurologist's clinical reasoning in movement disorders

Toward a better way for neurologists to onboard artificial intelligence: AI as assistant, monitor, coach, and teammate

Toward postCOVID-19 pandemic conditions: how artificial intelligence impacts movement disorders care and research

Neurologists and artificial intelligence: the two paths joining together

References

List of abbreviations

ADL   Activities of Daily Living

AI   Artificial Intelligence

CNN   Convolutional Neural Network

COVID-19   Coronavirus Disease 2019

CSF   Cerebrospinal Fluid

kNN   k-Nearest Neighbors

MDS   Movement Disorder Society

ML   Machine Learning

MPTP   1-Methyl-4-Phenyl-1,2,3,6-Tetrahydropyridine

NMS   Nonmotor Symptoms

PD   Parkinson's Disease

PPMI   Parkinson Progression Marker Initiative

RF   Random Forest

SVM   Support Vector Machine

TSRI   Thailand Science Research and Innovation

UPFRS   Unified Parkinson's Disease Rating Scale

WHO   World Health Organization

Introduction

Artificial Intelligence (AI) is one of the most exciting methodological shifts in our era, offering a huge potential to transform healthcare with humans and machines working together to provide better patient care. Although the term AI is fast becoming a new buzzword as it permeates our daily lives, credit must go to Alan Turing and others who, in the early 1940s, originally explored the possibility that machinery could show intelligent behavior. In his 1950 seminal paper, Computing Machinery and Intelligence, Turing began by proposing the question Can machines think? which developed into, in essence, Can machines do what we (as thinking entities) can do?, highlighting a fairly sharp line between the physical and intellectual capacities of man (Turing, 1950). Further to this point, he stated in his paper there was little point in trying to make a ‘thinking machine’ more human. In our interpretation, Turing's intention was probably not to replace humans with the machine, but instead to develop a learning machine in which rules were changed along the learning process, and the developing engineer might not know what was going on inside. Turing's idea sparked great interest in the field, with the term Artificial Intelligence being coined in 1956 to refer to the science and engineering of making intelligent machines, but its definition is continuously evolving with the latest referring to the collection of technologies that equip machines with higher levels of intelligence enabling them to perform various tasks such as perceiving, learning, problem-solving, and decision-making (Ahuja, 2019; Ganapathy, 2022). It was not until the 1990s that machine learning (ML), a particular form of AI which gives computers the ability to learn from experience and improve with outcomes without being explicitly programmed, was more widely utilized. Recent ML tools, such as support vector machines and recurrent neural networks allow clinician-scientists to leverage computational power to find patterns in data and build algorithms that are robust to data variation, leading to improved prognostic and diagnostic models in neurological disorders, including Parkinson's disease (PD) (Pedersen et al., 2020). The arrival of more powerful processors and ever-growing quantities of training data have further accelerated the growth of AI applications in healthcare, particularly with the development of deep learning methods, which aim to imitate the working of the human brain in processing vast amounts of information. We often picture an artificial neural network as having layers of nodes that are analogous to nuclei and connected by a series of mathematical weights like the synapses between them. These multiple nodes in a deep learning network learn by propagating information between multiple hidden network layers and give outputs according to a set of predefined rules that aids learning (Faust et al., 2018; Schmidhuber, 2015). The most recent advances of AI deep learning methods can even imitate the working of human brain in that the computer can rewire itself dynamically to take in new data as the brain does, enabling AI models to continue learning over time (Zhang et al., 2022).

While we continue to see rapid growth of AI applications in healthcare, there is a rising concern that AI will, in the future, outperform or even replace physicians (Ahuja, 2019). However, this concern is not equally shared among different specialties. Speculations that humans would soon be displaced by ML were initially focused on radiologists and anatomical pathologists, especially in the area of screening where machine accuracy has been found to be identical or even exceeding that of humans (Ardila et al., 2019; Esteva et al., 2017; Obermeyer & Emanuel, 2016). Neurology as a specialty is no exception. Increasingly, AI's ability to transform data into knowledge has disrupted at least two areas of neurology. First, ML has been shown and continues to improve diagnostic accuracy in many neurological conditions, with significant advances established in the field of diagnostic neuroimaging. One good example is deep learning algorithmic approaches that accurately identify head CT abnormalities requiring urgent attention (Chilamkurthy et al., 2018). These advances open up the possibility to use such algorithms to automate the triage process even in areas where neuroradiologists are lacking (Chilamkurthy et al., 2018). Second, ML will dramatically improve the ability of neurologists to establish a prognosis. While traditional prognostic models are restricted to only a handful of variables because humans must enter and tally the scores, data could instead be drawn directly from electronic health records, claim databases, activities of daily living (ADLs) or even monitoring outcomes, allowing models to use and integrate thousands of rich predictor variables. Early evidence from our work (RB), using ML to predict risk factors for falling based on daily activities of patients with PD, identified high-risk activities that involve cognitive attention and changes in vertical and lateral orientations (Panyakaew et al., 2021).

In our own view, as clinical neurologists with an interest in the use of technologies in the field of movement disorders, AI and neurologists are most potent when they combine their strength and cooperate (Bhidayasiri, 2021; Dzobo et al., 2020; Goetz, 2021). It should never be AI versus neurologists, as it is a symbiotic relationship between human expertise and predictive AI algorithms. The term Augmented intelligence refers to the assistive role of AI in advancing human capabilities and decision-making (Gennatas et al., 2020). An AI program can provide a decision or prediction after learning patterns from data, but the interpretation and real-world implementation of AI models requires human expertise (Blois, 1980; Pedersen et al., 2020). Therefore, neurologists must decide how AI will be best integrated into neurological clinical practice. Clearly, adopting new technologies into clinical practice can be a major challenge, and the more impact a technology has, the bigger that challenge is. AI may be perceived by some neurologists as particularly difficult to implement. To make it relevant to clinical practice, we begin this chapter by outlining the benefits of AI in the simulation of clinical reasoning—a process that all neurologists are used to and practice with their patients on a daily basis. The terms clinical reasoning, clinical judgment, and critical thinking are often used interchangeably, but they are not one and the same. While clinical reasoning is a cognitive and metacognitive process used for analyzing knowledge relative to a clinical situation or specific patient, clinical judgment is a little different, referring to the cognitive, psychomotor, and affective processes that are demonstrated through action and behaviors (Gruppen, 2017; Victor-Chmil, 2013). Critical thinking is the cognitive processes used for analyzing knowledge. Together, these processes lead to competent clinical practice. As part of the handbook of digital technologies in movement disorders, the focus of this chapter is on movement disorders, with most examples related to PD. Understanding the process of AI that is similar to clinical cognition should enable neurologists to get past their fears and build a trusting relationship with AI as a partner in the successful management of PD, ranging from the diagnosis, monitoring symptoms, making therapeutic decisions, detection of serious adverse events or complications, and identification of suitable candidates for device-aided therapies. Our message throughout this chapter is that to be a trusted partner, neurologists must onboard AI as their assistant, their colleague and teammate, and in certain situations their coach (Babic et al., 2020; Bhidayasiri, 2021).

Clinical reasoning in movement disorders: Neurologist's view

Interpreting symptoms and physical signs in movement disorders is usually not a binary judgment. Neurologists generally exercise their clinical judgment based on the information they receive, and pertinent information is likely to vary among individual patients and over time (Mandl & Bourgeois, 2017). The reasoning process involved in every clinical scenario can be demonstrated in the form of a funnel, with its wide start representing the reasoning process just before the first contact and the shrinking diameter reflecting the narrowing of possibilities with progressive acquisition of new information, culminating in the narrow conclusion or final diagnosis and management decision (Fig. 2.1) (Blois, 1980; Vishnu & Vinny, 2019). Let us imagine the scenario of a neurologist in a busy clinical practice being consulted on a 35-year-old man who presents with a 1-year history of progressive walking and speech difficulties. His main clinical complaints certainly create a wide possibility of diagnoses. Experienced neurologists will narrow down the possibilities when encountering this patient through a process of focused clinical interview and selected examination to aggregate the findings, followed by the selection of a pivot that determines the physician's categorical determination that this young man's complaints are due to parkinsonism, cerebellar, pyramidal or other system dysfunction. This anatomical step leads to the generation of a list of possible diagnoses, based specifically on the physician's mental skills educationally and experiencially acquired—that of comparing patterns and sequences to judge what information is relevant to achieving or approximating formal sets of diagnostic criteria. Trainees are often astonished at their professors' ability to sort their way through details, clear the confusion, and come up with a short list of differential diagnoses (Fig. 2.2). As more information becomes available from guided investigations and the passage of time, the possible causes are pruned down before arriving at a final diagnosis. Indeed, pattern recognition and selection of a pivot or constellation of key findings are identified as an essential step toward generating a differential diagnosis (Eddy & Clanton,