Introduction
The COVID-19 pandemic highlighted a central challenge in global health promotion: how to encourage and sustain protective behaviors on a population scale. The effectiveness of protective policies, such as social distancing, masking, and hand hygiene, depends entirely on broad public uptake.Reference Iezadi, Gholipour and Azami-Aghdash 1 Yet, achieving compliance requires more than simply disseminating health information.Reference Glanz, Rimer and Viswanath 2 It demands careful attention to the design of prevention strategies, specifically the delicate balance between voluntary appeals to individual responsibility and legally mandated restrictions on behavior. Understanding what drives compliance is critical not only for acute pandemic response but also for translating public health guidance into sustained, collective action.Reference Fishbein 3
Among these protective behaviors, reductions in mobility through social distancing have been central to policy interventions. Because mobility is measurable at scale, it provides a practical focus for comparative analysis. This focus highlights a broader tension in health promotion strategy, as countries have adopted markedly different approaches to achieve compliance.Reference Yan, Zhang and Wu 4 Evidence from diverse national experiences illustrates an ongoing debate over the relative effectiveness of voluntary versus mandated measures. In China, compulsory social distancing was highly effective in curbing transmission, while voluntary self-isolation emerged later and contributed little to reducing the spread.Reference Gupta, Simon and Wing 5 In the United States, by contrast, a substantial share of early mobility reductions occurred voluntarily before formal stay-at-home mandates were enacted.Reference Chudik, Pesaran and Rebucci 6 These contrasting experiences raise a vital question for global health policy: can voluntary action achieve comparable protective effects in developing country contexts, where structural, social, and economic conditions differ markedly from those in high-income nations?
From a health promotion perspective, voluntary and mandated social distancing represent alternative strategies for achieving compliance, each carrying distinct advantages and limitations. Voluntary measures rely on individuals responding to perceived risks and social norms.Reference Rosenstock, Strecher and Becker 7 However, they are inherently subject to externalities, as people may underappreciate the impact of their actions on the wider community.Reference Eichenbaum, Rebelo and Trabandt 8 Voluntary compliance can also be slow to materialize and is often constrained by economic or structural factors—barriers that are particularly salient in developing countries.Reference Zachary and Mobarak 9
Mandated measures, by contrast, can accelerate adherence and ensure a more uniform behavioral response. Yet, enforcement is rarely perfect, and strict mandates can trigger behavioral adaptation, where individuals compensate in ways that dilute the intended public health effect.Reference Sunstein 10 Crucially, mandates often produce severe unintended harms, including economic strain, exacerbated social inequities, and increased risks of domestic violence.Reference Bradbury-Jones and Isham 11 , Reference Usher, Bhullar and Durkin 12 Understanding these trade-offs frames the two approaches not as mutually exclusive, but as complementary prevention strategies. Thus, it is essential to empirically evaluate their relative effectiveness across different socioeconomic settings.
India offers a unique and highly relevant setting to empirically examine this dynamic. The Janta (public) Curfew (JC)—a unique, voluntary intervention—preceded the imposition of a nationwide mandated lockdown, providing a clear baseline for voluntary compliance before formal enforcement began. This sequential implementation allows for a natural comparison of how the same population responded to voluntary appeals versus mandated measures, controlling for cultural, economic, and social factors within a single country. Furthermore, as a developing nation, India presents structural conditions that are critically underrepresented in the existing literature. Analyzing this context offers valuable insights into how structural and behavioral constraints shape compliance with health promotion strategies. By leveraging this distinctive sequence of interventions, this analysis sheds light on the relative contributions of voluntary versus mandated approaches in promoting social distancing.
To achieve this, the study utilizes Google Community Mobility Reports (CMRs), 13 which provide anonymized, aggregated mobile data across various location types. Their high temporal resolution allows for a consistent, population-level assessment of behavioral change, ultimately guiding more flexible, context-specific health promotion interventions in resource-limited settings.
Methods
Study Design/Setting
This study examines mobility dynamics in India during February 15–May 31, 2020. The timeline includes 2 sequential interventions that provide a natural experiment for comparing voluntary and mandated social distancing: a 1-day voluntary public curfew (JC), 14 followed by a nationwide mandated lockdown. 15 On March 21, 2020, the Prime Minister urged citizens to observe a nationwide JC on March 22, lasting from 7:00 AM to 9:00 PM. The nationwide lockdown was implemented in multiple phases (Table 1), reflecting variations in policy stringency and permitted activities. May 31, 2020, was the final day of the national lockdown. Mobility data are available beginning February 15, 2020.
Definition and duration of the 10 analytical sub-periods
Table 1. Long description
Beginning at the top, the table has three columns labeled Dates, Phase, and Days. The first five rows list weekly periods before Janta Curfew: February 16 to February 22 as 5 weeks prior to J C for 7 days, February 23 to February 29 as 4 weeks prior to J C for 7 days, March 1 to March 7 as 3 weeks prior to J C for 7 days, March 8 to March 14 as 2 weeks prior to J C for 7 days, and March 15 to March 21 as 1 week prior to J C for 7 days. The sixth row is March 22 to March 24 labeled Janta Curfew for 3 days. The seventh row is March 25 to April 14 labeled Lockdown Phase 1 for 21 days. The eighth row is April 15 to May 3 labeled Lockdown Phase 2 for 19 days. The ninth row is May 4 to May 17 labeled Lockdown Phase 3 for 14 days. The tenth row is May 18 to May 31 labeled Lockdown Phase 4 for 14 days. Each phase is defined by its date range and duration in days.
Note: This table outlines the temporal windows used to evaluate the shift in mobility patterns across the different phases of the public health response.
Data
Outcome variables: Mobility trends are measured using Google Community Mobility Reports (CMRs), which provide daily, state- and union territory-level data across 6 location categories (retail and recreation [RRE], parks [PRK], transit stations [TST], grocery and pharmacy [GPH], workplaces [WRK], and residential [RSD]). Mobility at the 5 public locations is measured as the percentage change in total visitors from a baseline, defined as the median value for the corresponding day of the week during a 5-week reference period (January 3–February 6, 2020). For RSD, CMR mobility data reflects the change in time spent at home relative to the median value for the corresponding day of the week in the same reference period. Changes in mobility are a practical proxy for protective behavior, with declines in visits to 5 public locations and increases in time spent at home indicating greater adherence to social distancing.
Control: Daily COVID-19 case and mortality data at the state-/union territory-level are incorporated to account for regional variation in the epidemic’s intensity. These data provide essential controls to distinguish mobility changes driven by government interventions from those potentially influenced by local outbreak dynamics. State-level socioeconomic, demographic, and health infrastructure indicators are included as controls to account for heterogeneity across regions. These encompass factors such as population size and density, economic output, urbanization, and availability of health services. Further details on these controls, along with summary statistics, are provided in the Online Appendix.
Statistical Analysis
OLS regressions (Equation 1) are employed to estimate region-level (
$ s $
) mobility changes during 10 sub-periods (
$ p $
). The dependent variable
$ MO{B}_{s,t} $
is the percentage change in mobility on date
$ t $
.
$ {D}_{p(t)} $
is the sub-period dummies (equaling 1 if
$ t $
falls in sub-period
$ p $
). Coefficients (
$ {\beta}_p $
) on pre-JC and JC dummies capture voluntary behavior adjustments, whereas coefficients on lockdown sub-periods dummies capture additional changes associated with mandated restrictions.
To reduce confounding, the regressions include regional controls that capture both pandemic dynamics (
$ PA{N}_{s,t-1} $
) and structural characteristics (
$ {X}_s $
). These variables help account for regional differences in exposure risk and preparedness that might otherwise drive mobility changes. Pandemic controls comprise measures of local epidemic intensity (1-day lag of log(1 + cumulative cases per million) and log(1 + deaths per million)). Structural controls include population size, population density, net state domestic product per capita, urbanization rate, and proxies for local health infrastructure.
Regional fixed effects
$ {\sigma}_s $
absorb time-invariant regional heterogeneity, while day-of-week fixed effects
$ {\tau}_t $
capture weekly seasonality. Standard errors are clustered by date to account for common shocks (e.g., festivals, holidays) that may influence mobility independently of pandemic dynamics.
Estimating voluntary share: The study design helps separate voluntary and mandated changes in mobility. The assumption is that reductions in mobility during the JC can be interpreted as evidence of voluntary compliance, with reductions during the lockdown phases including the effects of mandated restrictions. Therefore, for any particular CMR location-type, the upper bound of the voluntary share is defined as the ratio of the JC effect (
$ \mid {\beta}_{JC}\mid $
) to the maximum effect observed in any lockdown phase
$ \left({\max}_{p\;\epsilon\;\left\{{P}_1,{P}_2,{P}_3,{P}_4\right\}}|{\beta}_p|\right) $
, where
$ {P}_i $
denotes the
$ i\mathrm{th} $
lockdown phase. In addition, 3 counterfactual benchmarks provide lower-bound estimates: (i) a continuation of the pre-Curfew trend in mobility, based on changes happening in the week immediately prior, (ii) a continuation of the pre-Curfew trend based on changes happening in the 2 weeks preceding JC, and (iii) a conservative version of the trend in (i) (scaled to 75%). These benchmarks reflect the fact that individuals were already practicing prophylactic adjustments as infection risks rose. As this behavior change preceded both JC and mandates, they are inherently voluntary. Together, the upper- and lower-bound estimates offer a transparent range for the share of the lockdown effect that can reasonably be attributed to voluntary behavior.
Results
Descriptive Trends
Figure 1 shows national mobility patterns (lines) across 6 Google CMR categories, plotted with cumulative COVID-19 cases (bars). Panel A covers the complete study period; Panel B zooms in around the JC.
Trends in cumulative COVID-19 cases and mobility across distinct location categories.
Note: The vertical green line indicates the Janta Curfew (March 22, 2020), and the red line indicates the initiation of the national lockdown. Bars represent cumulative cases (left-hand axis); lines represent percentage change in mobility from baseline (right-hand axis). Mobility change is calculated using a 5-week baseline (January 3–February 6, 2020), adjusted for day-of-the-week effects. 13
Figure 1. Long description
Panel A at the top displays cumulative COVID-19 cases as blue vertical bars (left y-axis, 0 to 50000) and percent change in mobility from baseline as six colored lines (right y-axis, -100 to 50). The x-axis spans February 15 to May 5. Mobility categories are retail and recreation (red), parks (orange), transit stations (yellow), grocery and pharmacy (light orange), workplaces (brown), and residential (green). The green vertical line marks Janta Curfew on March 22, and the red vertical line marks the start of the national lockdown. Before March 22, mobility lines are near baseline (0 percent change). After March 22, all mobility categories except residential drop sharply to around -50 to -100 percent, while residential increases to about 30 percent above baseline. Cumulative cases remain low until late March, then rise steeply after the lockdown. Panel B below focuses on March 1 to March 28, with cumulative cases (left y-axis, 0 to 1000) and mobility (right y-axis, -100 to 50). The same color scheme and vertical lines are used. Mobility drops sharply after Janta Curfew, with residential rising, and cumulative cases begin to increase after the interventions. Legends for both panels identify each mobility category by color.
Three features stand out. First, mobility dropped sharply on March 22 during the JC, indicating widespread voluntary compliance. Second, further declines began on March 25 with the nationwide lockdown. In line with these declines, RSD rose, reflecting more time at home. Third, as lockdowns continued, mobility partially rebounded but remained below pre-JC levels through May.
These reductions occurred alongside rising case counts, suggesting that voluntary risk perception likely reinforced government restrictions. Some short-term dips (e.g., March 10, Holi) were unrelated to COVID-19. Date fixed effects in the regression models absorb such deviations.
Regression Estimates
Figure 2 plots estimated coefficients for each location across 9 sub-periods, relative to 5 weeks before the JC. Before mid-March, mobility was stable, except for modest increases for WRK. Significant declines emerged the week before JC, especially in non-essential locations (RRE and PRK down 23–35 pp) and more modestly in essential ones (GPH and WRK down 2–21 pp). Time spent at home increased (RSD up 4 pp). These early changes underscore the voluntary mechanism, which was in operation long before any mobility mandates or directives were placed.
Estimated changes in mobility relative to baseline across pre-curfew, Janta Curfew, and lockdown phases.
Note: Point estimates and 95% CIs correspond to coefficients on sub-period dummies
$ {D}_{p(t)} $
in the regression model (Equation 1). The vertical green and red lines denote the Janta Curfew and the start of the lockdown, respectively.
Figure 2. Long description
Top row, left to right: Retail and recreation panel shows mobility near zero before curfews, dropping steeply below minus sixty at Janta Curfew and lockdown, then slowly rising but remaining negative through P4. Parks panel follows a similar pattern, with a steep drop at curfews and partial recovery. Transit stations panel shows a sharp decline from near zero to below minus sixty at curfews, then a gradual increase but not returning to baseline. Bottom row, left to right: Grocery and pharmacy panel shows a moderate decline before curfews, a sharp drop at curfews, then a steady increase through P4 but still below baseline. Workplaces panel shows a steep drop at curfews, then a gradual increase but remaining negative. Residential panel shows a sharp increase at curfews, peaking above plus twenty, then stabilizing and slowly declining but staying above baseline. X-axes are labeled with weeks before curfew (4wk, 3wk, 2wk, 1wk, J C), and lockdown phases (P1 to P4). Y-axes are labeled with percent change from baseline, with a dashed line at zero. Vertical green line marks Janta Curfew, red line marks lockdown start. Each panel uses a distinct color for its line: red for retail and recreation, red for parks, orange for transit stations, yellow for grocery and pharmacy, orange for workplaces, green for residential.
During the JC and subsequent lockdowns, the sharpest drops usually occurred in RRE, TST, and PRK. During the JC, RRE fell 67.4 (95% CI: 66.8–68.0) pp, TST 64.0 (63.3–64.7) pp, PRK 41.0 (95% CI: 40.1–41.9) pp, WRK 30.7 (95% CI: 29.9–31.4) pp, and GPH 58.7 (95% CI: 58.0–59.4) pp. Time at home rose 16.3 (95% CI: 16.0–16.6) pp. These reductions persisted through later lockdown phases, particularly in non-essential categories, though they gradually moderated over time. WRK and GPH saw milder and temporary effects, while RSD remained elevated before also tapering. Nearly all estimates are statistically significant at the 95% level.
The online appendix contains the complete regression results.
Changes Attributable to Voluntary Channel
Table 2 summarizes upper- and lower-bound estimates of voluntary contributions to mobility reductions. Column (1) shows the maximum decline during lockdowns, column (5) the change during JC, and their ratio in column (9) gives the implied upper bound.
Estimated changes in mobility at the Janta Curfew and the calculated voluntary share of mobility reduction
Table 2. Long description
From top to bottom, the table lists six location types: Retail and recreation, Parks, Grocery and pharmacy, Workplaces, Transit stations, and Residential. For each, columns show maximum change during lockdown, voluntary decline at Janta Curfew as per trend (for 1 week, 0.75 times 1 week, and 2 weeks), Janta Curfew value, and voluntary decline at Janta Curfew as per trend (rate of daily change for 1 week, 0.75 times 1 week, and 2 weeks), plus Janta Curfew rate. Retail and recreation: maximum change minus 79.4, voluntary declines minus 54.0, minus 49.3, minus 46.0, Janta Curfew minus 67.4, rates 0.68, 0.62, 0.58, Janta Curfew rate 0.85. Parks: maximum change minus 43.2, voluntary declines minus 38.6, minus 34.7, minus 30.7, Janta Curfew minus 41.0, rates 0.89, 0.80, 0.71, Janta Curfew rate 0.95. Grocery and pharmacy: maximum change minus 59.6, voluntary declines minus 27.3, minus 25.7, minus 25.1, Janta Curfew minus 58.7, rates 0.46, 0.43, 0.42, Janta Curfew rate 0.98. Workplaces: maximum change minus 29.1, voluntary declines minus 9.7, minus 7.8, minus 9.4, Janta Curfew minus 30.7, rates 0.33, 0.27, 0.32, Janta Curfew rate 1.05. Transit stations: maximum change minus 63.4, voluntary declines minus 45.1, minus 40.9, minus 38.4, Janta Curfew minus 64.0, rates 0.71, 0.65, 0.61, Janta Curfew rate 1.01. Residential: maximum change 18.0, voluntary declines 6.2, 5.6, 6.1, Janta Curfew 16.3, rates 0.34, 0.31, 0.34, Janta Curfew rate 0.90. Clustered standard errors are shown in parentheses below each main value. Panel A estimates are statistically significant at p less than .01.
Note: Estimates in Panel A are statistically significant at
$ p<.01 $
. Clustered standard errors in parentheses.
Voluntary behavior accounted for a large share of observed reductions. For WRK and TST, the JC declines exceeded those in any lockdown phase, implying upper bounds above 100%. RRE showed the smallest voluntary share (~85%), with other locations in between.
Figure 3 illustrates the assumptions behind lower-bound estimates. The solid line depicts estimated coefficients; dashed lines show counterfactual trajectories at JC based on pre-JC trends—one continuing the prior slope, the other adjusted to 75% of that slope. The estimated JC coefficient remains well below both (above both for RSD), indicating that the dip in mobility cannot be explained solely by a reasonable extrapolation of ongoing trends. Instead, it indicates that though voluntary responses were already active, the initiation of JC had a profound effect on people’s behavior.
Estimated voluntary mobility reduction during the Janta Curfew.
Note: Point estimates (solid dots) correspond to coefficients on sub-period dummies
$ {D}_{p(t)} $
in the regression model (Equation 1). The vertical green line denotes the Janta Curfew. Dashed lines represent counterfactual trajectories calculated based on pre-Janta Curfew trends to isolate the voluntary component of mobility change.
Figure 3. Long description
Top row, left to right: Retail and recreation, Parks, and Transit stations panels all show y-axes from minus 75 to plus 25, x-axes from 4 weeks before to J C. Each shows a flat or mild decline until 1 week before J C, then a steep drop after the green vertical line at J C. Dashed lines and open markers indicate counterfactuals based on pre-J C trends. Middle row: Grocery and pharmacy and Workplaces panels show similar patterns, with a sharp decline after J C. Residential panel (far right) has a y-axis from 0 to 20 and shows a marked increase after J C. All panels use solid lines for observed data, dashed for counterfactuals, and include a legend for counterfactual calculation methods. The green vertical line consistently marks J C across all panels.
Columns (2)–(4) of Table 2 report predicted JC declines under these counterfactuals, and columns (6)–(8) show corresponding voluntary shares relative to lockdown maxima. The voluntary share is highest for PRK (71–89%) and lowest for WRK (27–33%), with TST at 61–71%. For RSD, which moves oppositely, voluntary shares are smaller (31–34%).
Overall, these estimates confirm that voluntary behavior played a vital role in reducing mobility. Even under conservative assumptions, voluntary actions explain 27–71% of the maximum lockdown effects, demonstrating that self-initiated responses were substantial and measurable well before formal mandates.
Discussion
This study assesses how voluntary behavior and mandated restrictions shaped mobility during India’s early COVID-19 outbreak, offering new evidence on their relative effectiveness. Four main insights emerge.
-
(a) Across mobility categories, voluntary behavior played a substantial role in reducing movement. Mobility began declining nearly two weeks before formal government directives, consistent with rising awareness and precautionary behavior. Conservative estimates suggest that 27–71% of lockdown-era mobility reductions can be attributed to voluntary actions, with non-mandated appeals (JC) sometimes matching or exceeding declines achieved during lockdowns. This finding extends existing evidence on pandemic behavioral responsesReference Gupta, Simon and Wing 5 , Reference Fenichel, Kuminoff and Chowell 16 –Reference Lee and You 19 to a low- and middle-income setting. Discussions of voluntary versus mandated compliance in these settings have been largely anecdotal.Reference Zachary and Mobarak 9 , Reference Loayza 20 The early declines challenge the assumption that stringent mandates are always the principal driver of protective behavior and highlight the salience of community and individual initiative. As in high-income countries,Reference Gupta, Simon and Wing 5 , Reference Andersen 21 –Reference Spelta, Flori and Pierri 24 non-mandated interventions, such as public appeals and risk communication, can trigger meaningful behavioral change, reducing reliance on disruptive lockdowns while supporting effective containment in developing economies.
-
(b) The effects of voluntary and mandated measures differed sharply across locations. Non-essential places, such as retail areas and parks, recorded the steepest declines. Essential locations, such as grocery stores, pharmacies, and workplaces, showed milder contractions. Voluntary reductions were also more pronounced at transit stations, suggesting people curtailed travel even before mandates took effect, despite the heavy restrictions on transportation during lockdowns. These patterns confirm that not all forms of mobility are equally elastic. In line with international findings,Reference Maloney and Taskin 25 , Reference Weill, Stigler and Deschenes 26 people cut non-essential trips first while maintaining those linked to livelihood. Yet India’s unusually high voluntary reductions in transit suggest that even essential mobility can respond to risk perception when infection fears are acute.Reference Padmakumar and Patil 27 The policy implication is that governments can combine targeted restrictions on non-essential movement with communication strategies that encourage voluntary restraint in essential travel.
-
(c) In most cases, mandated lockdowns did not generate greater reductions than voluntary behavior. As in developed countries,Reference Gupta, Simon and Wing 5 , Reference Andersen 21 –Reference Pullano, Valdano and Scarpa 23 mobility declines often preceded formal restrictions, and in some locations, such as workplaces and transit, lockdown effects were smaller than those during the JC. These findings challenge the assumption that strict orders are always the most potent mechanism for reducing mobility. In developing contexts, where livelihood costs are high, mandates may be more valuable for sustaining compliance than for deepening it.Reference Bavel, Baicker and Boggio 18 , Reference Herlitz, MacIntyre and Osborn 28 , Reference Porat, Nyrup and Calvo 29 Sequencing thus matters: voluntary appeals can trigger compliance with health directives, while mandates can consolidate and stabilize those shifts once underway.
-
(d) Finally, the effect of lockdowns in keeping people at home weakened over time. Individuals adapted to prolonged restrictions, balancing compliance with social and economic needs. This dynamic supports evidence that public health interventions are rarely static, with people adjusting as circumstances evolve.Reference Bavel, Baicker and Boggio 18 , Reference Bardosh, Figueiredo and Gur-Arie 30 , Reference Schmid, Böhm and Das 31 The value of lockdowns, therefore, lies more in buying time than in sustaining reductions indefinitely. They cannot be employed as one-off solutions and need to be integrated into adaptive strategies (e.g., clear communication, support for essential needs, calibrated easing) to maintain compliance without creating counterproductive incentives. In short, the success of lockdowns depends less on their severity than on their integration within a flexible, evolving public health framework.
Limitations
The study has several limitations.
The first concerns the population coverage of mobile-based mobility data. These datasets disproportionately capture individuals with reliable internet access and smartphones, underrepresenting poorer and rural populations who often lack such access.Reference Grantz, Meredith and Cummings 32 , Reference Parker, Fraser and Abeler-Dörner 33 This bias is particularly relevant in India, where connectivity remains uneven across gender,Reference Scott, Shinde and Ummer 34 education,Reference Nidhi, Kamath and Ilavarasan 35 and minority status.Reference Nidhi, Kamath and Ilavarasan 35 Because these groups are also more vulnerable to the economic and health consequences of COVID-19, their underrepresentation may bias inferences about mobility and compliance. Future research could address this gap by combining smartphone-based measures with alternative sources (e.g., travel surveys, transport ticketing, and electricity consumption) to achieve broader coverage.
The second limitation involves the overlap between voluntary behavioral responses and government mandates. In many cases, mobility declined shortly before policy announcements, making it difficult to isolate the causal impact of voluntary versus mandated actions. The lower- and upper-bound strategy used here helps bracket the possible contribution of each, but cannot eliminate simultaneity bias. Hence, the estimates should be viewed as indicative rather than definitive, particularly when voluntary and policy-driven responses occurred in close succession. Future studies could exploit settings where these responses were more clearly staggered across time.
Finally, the analysis focuses exclusively on India, a country marked by large regional diversity and complex social and economic structures.Reference Sinha 36 While this provides a rich testing ground, the findings may not generalize to contexts with different institutional capacities, social norms, or enforcement mechanism.Reference Irina and Rose 37 –Reference Giménez-Nadal, Molina and Velilla 40 For instance, countries with stronger state capacity, higher/lower trust in government, or distinct urban–rural compositions might exhibit different balances between voluntary and mandated behavior. Comparative work using harmonized mobility data across developing economies could help clarify how contextual factors shape the relative effectiveness of voluntary appeals and mandates.
Conclusion
This study shows that voluntary behavior substantially reduced mobility during India’s early COVID-19 outbreak. By quantifying the voluntary channel, this analysis provides evidence that policy design can strategically leverage intrinsic behavioral responses, rather than relying solely on coercive measures, to achieve public health objectives. Future work should test these dynamics in other developing and epidemic settings.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/dmp.2026.10389.
Data availability statement
The data and code used in this research can be obtained by contacting the corresponding author. In addition, these files have been uploaded to the corresponding author’s website; access details and a direct link are provided in the Supplementary Material.
Author contribution
RS led the study conceptualization and methodology, performed the data curation and formal analysis, and was responsible for drafting and revising the manuscript.
Funding statement
The research has not received any funding from any institution.
Competing interests
The author has no relevant financial or non-financial interests to disclose.
Ethical approval
None sought as no human participants were involved.
