Evaluating the Effectiveness of AI- Powered Physical Therapy Intervention for Patients with Parkinsons Disease

Parkinson’s disease (PD) is a chronic, progressive neurodegenerative disorder that affects millions of individuals worldwide and represents one of the most challenging conditions within neurorehabilitation. Characterized by bradykinesia, rigidity, resting tremor, and postural instability, PD significantly impairs functional mobility and quality of life ¹. The global prevalence of PD has been rising steadily, attributed partly to increased life expectancy and improved diagnostic accuracy ². As the burden of PD continues to grow, the need for effective, accessible, and evidence?based rehabilitation strategies becomes increasingly urgent ³.

Conventional physiotherapy remains a cornerstone of PD management, targeting motor impairments, gait dysfunction, balance deficits, and functional limitations ⁴. Physiotherapy interventions such as gait training, cueing techniques, strengthening exercises, and balance rehabilitation have demonstrated benefits in improving mobility, reducing fall risk, and enhancing independence ⁵. However, traditional therapist?supervised rehabilitation often faces limitations related to session frequency, patient adherence, variability in exercise execution, and limited access to specialized care—particularly for individuals living in remote or underserved regions ⁶. These challenges highlight the need for innovative rehabilitation models that can complement conventional therapy and provide continuous, adaptive support ⁷ .

Recent technological advancements have introduced artificial intelligence (AI)–powered physical therapy systems capable of transforming the landscape of neurorehabilitation⁸. AI?based rehabilitation platforms leverage real?time motion analysis, predictive modeling, and adaptive feedback to deliver personalized exercise programs tailored to the unique needs and performance of each patient ⁹ . These systems can monitor movement precision, detect errors, make real?time adjustments, and track progress longitudinally, offering a level of consistency and customization that surpasses traditional approaches¹⁰. Moreover, AI?enabled telerehabilitation tools allow individuals with PD to engage in structured therapy sessions at home, potentially improving adherence and reducing the logistical burdens associated with in?person visits¹¹.

The integration of AI into PD rehabilitation aligns with the broader shift toward digital health solutions and precision medicine ¹². Intelligent rehabilitation platforms can analyze large datasets collected from sensors, cameras, and wearable devices, enabling detailed insight into patient performance patterns ¹³. Through machine learning algorithms, these systems can identify subtle fluctuations in motor behavior, predict therapy needs, and optimize intervention plans more effectively than manual methods ¹⁴. In PD, where motor symptoms fluctuate throughout the day and vary across disease stages, such adaptive technologies may provide meaningful clinical advantages¹⁵.

Despite the promising potential of AI?powered rehabilitation, current evidence on its clinical effectiveness in Parkinson’s disease remains limited ¹⁶. While preliminary studies have explored the use of virtual reality, sensor?based training, and gamified platforms in PD therapy, research specifically focused on AI?guided physiotherapy is still evolving ¹⁷. Questions remain regarding the magnitude of improvement achievable with AI?assisted interventions compared with standard physiotherapy, as well as the feasibility, usability, and long?term adherence associated with these technologies ¹⁸. Moreover, clinicians require high?quality evidence from randomized controlled trials to confidently integrate AI?based tools into routine PD management ¹⁹.

This study aims to address these gaps by evaluating the effectiveness of an AI?powered physical therapy intervention on motor function, gait parameters, balance, and quality of life in patients with mild to moderate Parkinson’s disease²⁰. By comparing outcomes between individuals receiving AI?guided rehabilitation and those undergoing conventional therapist?supervised physiotherapy, the study seeks to contribute robust evidence regarding the clinical value of intelligent digital tools in PD rehabilitation²¹. With the growing global emphasis on technology?enabled healthcare, findings from this research may inform future rehabilitation models and support the integration of AI?driven systems into physiotherapy practice²².

METHODOLOGY
1. Population

The target population for this study includes individuals diagnosed with Idiopathic Parkinson’s Disease (PD) residing in Dehradun, Uttarakhand, India. Participants belong to Hoehn and Yahr Stages I–III, representing mild to moderate PD where rehabilitation therapies remain clinically effective. This population is selected because early to middle stages of PD demonstrate better responsiveness to motor learning strategies, physical training, and technology-assisted interventions.

2. Source of Participants

Participants will be recruited from multiple hospitals and rehabilitation centers in Dehradun, including:

• Government Doon Medical College & Hospital
• Max Super Specialty Hospital
• Himalayan Hospital (Jolly Grant)
• Synergy Physiotherapy & Rehabilitation Center
• Guru Nanak College of Paramedical Sciences Physiotherapy OPD

These centers regularly treat PD patients, ensuring appropriate participant availability.

3. Sample

A total sample of 40 participants (20 per group) will be included. Power analysis supports that this sample size can detect meaningful differences in outcomes. Participants will be divided into:

• Experimental Group (AI-based Therapy): 20 participants
• Control Group (Conventional Physiotherapy): 20 participants

4. Place of Study

The study will be conducted primarily at the Physiotherapy Department and Research Laboratory of Guru Nanak College of Paramedical Sciences & Hospital, Dehradun, where assessment equipment and AI-based rehabilitation tools are available.
5. Research Design

A Randomized Controlled Trial (RCT) design will be used. This pre-test/post-test experimental design allows comparison between AI-powered physiotherapy and conventional physiotherapy. Random allocation will be performed using computer-generated randomization.

6. Selection Criteria

6.1 Inclusion Criteria

• Diagnosed with Idiopathic Parkinson’s Disease
• Hoehn and Yahr Stages I–III
• Age 45–75 years
• Stable medication for 4 weeks
• Able to walk independently
• Cognitively able (MMSE ≥ 24)
• Willing to participate and provide informed consent

6.2 Exclusion Criteria

• Hoehn and Yahr Stages IV–V
• Severe cognitive impairment or psychiatric illness
• History of stroke, TBI, or major neurological diseases
• Musculoskeletal limitations restricting mobility
• Uncontrolled cardiac or respiratory conditions
• Severe vision/hearing impairments
• Participation in experimental programs within last 3 months

7. Sampling Technique

Purposive sampling will be used to identify eligible PD patients from the selected hospitals. After screening, participants will be randomized into two groups using simple random sampling (computer-generated numbers). Allocation concealment will be maintained using sealed opaque envelopes.

8. Variables

Independent Variable:

• Type of Physiotherapy Intervention (AI-powered vs. Conventional)

Dependent Variables:

• Motor function (UPDRS-III)
• Balance (Berg Balance Scale)
• Functional mobility (Timed Up and Go Test)
• Gait parameters (speed, stride length)
• Quality of life (PDQ-39)

9. Instrumentation

• Unified Parkinson’s Disease Rating Scale (UPDRS-III)
• Berg Balance Scale (BBS)
• Timed Up and Go Test (TUG)
• PDQ-39 Questionnaire
• AI-based Motion Tracking System (wearable sensors, cameras)
• Gait analysis tools
• Stopwatch, cones, therapy mats

Intervention Protocol (8 Weeks, 5 Days/Week, 45–60 Minutes/Session)

Experimental Group: AI-Powered

Physiotherapy

Therapy includes real-time motion capture, adaptive correction, error detection, and personalized progression.

Components:

1. Warm-Up (5 min)
2. AI-Guided Motor Training (20 min)

• Reaching, coordination, object manipulation, sequencing tasks

3. AI-Based Gait Training (15 min)

• Step initiation, stride training, obstacle negotiation

4. Balance Training (10 min)

• Weight shifting, virtual balance tasks, dynamic stability challenges

5. Cool Down (5 min)

• AI provides visual, auditory, and performance-based feedback throughout.

Control Group: Conventional Physiotherapy

Components:

1. Warm-Up (5 min)
2. Mobility Training (15 min)

• Sit-to-stand, heel-toe raises, arm swing training

3. Gait Training (15 min)

• Cue-based walking (auditory/metronome cues)
• Treadmill and overground walking

4. Balance Training (10 min)

• Static/dynamic balance, tandem stance, reaching tasks

5. 5. Cool Down (5 min)

Flow Chart of Procedure

RESULTS

Statistical Analysis

The following statistical procedures were performed for all outcome measures:

• Descriptive statistics (Mean and Standard Deviation)
• Shapiro–Wilk test for normality
• Paired t-test for within-group comparisons
• Independent t-test for between-group comparisons
• Significance level set at p < 0.05
• Effect size computed using Cohen’s d

Table 1: UPDRS-III Raw Scores

Table 2: Berg Balance Scale Raw Scores

Table 3: Timed Up and Go (TUG) Raw Scores

Table 4: Gait Speed Raw Scores

Table 5: PDQ-39 Quality of Life Raw Scores

Explanation: UPDRS (Experimental Group) visually demonstrates the trend for pre- and post-intervention measurements. A downward shift indicates improvement in UPDRS and TUG, while an upward shift represents improvement in BBS, gait speed, and quality of life.

Explanation: UPDRS (Control Group) visually demonstrates the trend for pre- and post-intervention measurements. A downward shift indicates improvement in UPDRS and TUG, while an upward shift represents improvement in BBS, gait speed, and quality of life.

Explanation: TUG (Experimental Group) visually demonstrates the trend for pre- and post-intervention measurements. A downward shift indicates improvement in UPDRS and TUG, while an upward shift represents improvement in BBS, gait speed, and quality of life.

Explanation: Gait Speed Comparison visually demonstrates the trend for pre- and post-intervention measurements. A downward shift indicates improvement in UPDRS and TUG, while an upward shift represents improvement in BBS, gait speed, and quality of life.