Advancing Mobile Health Accessibility Through Data-Driven Modeling

Advancing Mobile Health Accessibility Through Data-Driven Modeling Siva Kumar Chintham Prabhav Services Inc Texas, USA Darshan Mohan Bidkar University of Texas Dallas Texas, USA Vishnu Ramineni Apollo 24/7 Telangana, India Nitin Saksena Albertsons Companies Inc California, USA Abstract Older adults represent a rapidly growing segment of mobile health (mHealth) application users, yet the interaction modalities dominant in commercial health applications fine-gesture touch, small-target navigation, and high-density information layouts are systematically misaligned with the perceptual, motor, and cognitive changes associated with ageing. This paper presents the Ageing-Adaptive mHealth Design Framework (AADF), a formal model that weights accessibility feature contributions using a Geriatric Usability Impact Score (GUIS) derived from age-stratified usability dimensions: visual acuity decline, motor precision loss, and working memory reduction. A controlled usability evaluation was conducted with 64 participants aged 60–85 across three prototype configurations: Baseline (BL, standard commercial interface), Partially Accessible (PA, WCAG 2.1 Level AA), and AADF-instrumented (AADF), across six standardised mHealth task types. AADF achieves a mean task completion rate (TCR) of 89.4% compared to 41.8% for BL across all tasks and age subgroups a 47.6 percentage-point improvement. Voice interaction reduced vital-sign entry time by 52% for participants aged 75–85. Progressive font scaling reduced reading-error rates by 44% for participants with selfreported low vision. Three structural design patterns Stepped Navigation, Persistent Context Anchors, and Modality-Fallback Sequences are formalised and evaluated, demonstrating collectively that navigation-complexity reduction accounts for 38% of the total AADF TCR improvement. The paper argues that age-specific mHealth accessibility requires a design philosophy extending beyond WCAG compliance into geriatric interaction engineering. Keywords Accessibility, Data Modelling, Voice interaction, Adaptive interface. 1 Introduction By 2030, adults aged 60 and over are projected to constitute 1.4 billion of the global population [1], a demographic with elevated chronic disease burden, higher healthcare utilisation, and correspondingly greater mHealth application de- Balakrishna Pothineni Virtusa Corporation Texas, USA Prema Kumar Veerapaneni Cognizant Technology Services Texas, USA pendence for medication management, vital-sign monitoring, and telehealth participation [3]. Commercial mHealth applications are designed primarily for users aged 25–45, with interaction paradigms pinch-to-zoom, small touch targets, scrolling-heavy information architecture that conflict with the physiological and cognitive changes of ageing [4]. The resulting usability gap is not a peripheral accessibility concern; it is a clinical access deficit: an older adult who cannot reliably complete a medication logging task in a health management application faces direct medication adherence risk. The WCAG 2.1 standard [6] provides the canonical accessibility specification, but its criteria were not derived from ageing-specific usability evidence. Criterion pass/fail compliance does not translate to usable interaction for older adults when, for instance, a 44 sp font satisfies Level AA but remains unreadable for a user with age-related contrast sensitivity reduction, or when keyboard operability is technically present but the focus-traversal sequence is incompatible with older adults’ navigation cognition models [4, 27]. Prior mHealth accessibility research for older adults is fragmented: studies of font scaling in isolation [8], touchtarget sizing [10], and voice interface adoption [16] have not been synthesised into a unified design framework with quantitative feature-weighting. No formal model exists that maps ageing-specific usability dimensions to mHealth feature effectiveness, or that structures the resulting design guidance as deployable architectural patterns [31]. Contributions: (1) The Ageing-Adaptive mHealth Design Framework (AADF), a formal feature-weighted usability model incorporating the Geriatric Usability Impact Score (GUIS). (2) Controlled evaluation of AADF across 64 participants aged 60–85 across three prototype configurations and six task types. (3) Formalisation and evaluation of three structural design patterns for older-adult mHealth navigation complexity reduction. 2 Literature Review 2.1 Ageing, Perception, and Mobile Interaction Fisk et al. [4] provided the foundational synthesis of cognitive ageing and technology design, establishing that age-related changes to visual acuity, contrast sensitivity, working memory capacity, and psychomotor speed are non-uniform across individuals and interact with technology complexity in non-additive ways. Darroch et al. [8] demonstrated that font size preferences in older mobile users diverge substantially from standard mobile type specifications[14], with users aged 65–80 preferring base sizes of 16–20 pt 50% larger than default iOS and Android system fonts at the time. Critically, their study found that merely increasing font size without adjusting line-height and inter-character spacing reduced readability by creating layout overflow artifacts that disrupted reading flow a finding with direct implications for dynamic font scaling implementation[32]. 2.2 Touch Target Design and Motor Impairment Park and Han [10] evaluated touch target performance across age groups and found that target acquisition error rates for adults aged 65+ were 3.4 times higher than for adults aged 25–34 at standard 7 mm target sizes, converging toward parity only at target sizes of 12 mm or greater. Wobbrock et al. [12] formalised the interaction between motor capability decline and target size as an ability-based design imperative interface designers should specify target dimensions relative to the expected capability distribution of the user population rather than the average user. In mHealth contexts, where critical interactions include medication dose confirmation and alert acknowledgement, motor-precision barriers constitute direct clinical risk. 2.3 Voice Interaction for Older Adults in Healthcare Bickmore et al. [16] demonstrated that conversational agent interfaces reduce health literacy barriers and improve medication adherence engagement for older and lower-literacy patients, with particular benefit for patients aged 65+ who reported lower confidence with touch-based interaction. However, their study examined voice as an isolated modality in a fixed clinical setting; the transition to mobilenative voice interaction introduces additional challenges including ambient noise robustness, wake-word activation reliability, and the cognitive demand of managing voice interface state awareness without visual feedback each of which requires mobile-specific design specification[14, 25]. 2.4 Navigation Complexity and Cognitive Load Nielsen [18] established the foundational principle that navigation depth beyond three levels induces disorientation in the majority of users, a finding subsequently quantified in the context of older adult web interaction by Leporini and Paternò [19], who found that navigation depth exceeding two levels increased task abandonment rates in older adults by 62%. The mechanism is working memory: multistep navigation requires maintenance of positional context across steps, which imposes working memory demands that conflict with age-related working memory reduction documented by Fisk et al. [4]. No prior work has formalised this relationship into deployable mHealth design patterns or quantified the impact of specific navigation architecture choices on older-adult task completion [21, 28]. 3 Methodology 3.1 Participant Sample 64 participants were recruited across three age strata: Young-old (60–69, n = 24), Middle-old (70–79, n = 24), and Old-old (80–85, n = 16). Inclusion criteria: selfreported smartphone use ≥3 times/week; no diagnosed dementia; corrected visual acuity sufficient for independent daily activities. Exclusion criteria: severe upper-limb motor impairment precluding any touch interaction; moderate or severe cognitive impairment [22, 23]. 3.2 Prototype Configurations Three mHealth application prototypes were developed targeting medication management, appointment scheduling, and vital-sign logging: BL (Baseline): Standard commercial mHealth interface, no accessibility modification[17]. PA (Partially Accessible): BL with WCAG 2.1 Level AA corrections (contrast, target sizing, labels). AADF: Full ageing-adaptive architecture with font scaling, voice pipeline, extended touch targets, and three design patterns[2]. 3.3 Task Battery Six mHealth tasks were evaluated per prototype: T1 medication log entry, T2 appointment booking, T3 vital-sign entry, T4 clinician message review, T5 medication reminder setup, T6 test result review. Primary metrics: TCR, Task Completion Time (TCT), Error Rate (ER). Secondary: NASA-TLX cognitive load subscale[5]. Table 1: GUIS Dimension Weights λd by Age Stratum Age Stratum Young-old (60–69) Middle-old (70–79) Old-old (80–85) 3.4 Geriatric (GUIS) λv (Visual) Table 2: Task Completion Rate (%) by Prototype and Age λm (Motor) λc (Cognitive) Stratum 0.28 0.33 0.38 Usability 0.34 0.38 0.41 0.38 0.29 0.21 Impact Score The GUIS quantifies the expected usability contribution of accessibility feature f for user capability profile a: X GUIS(f, a) = λd · eff(f, d) · (1 − ad ) (1) d∈D where D = {visual, motor, cognitive}, a = [av , am , ac ]T ∈ [0, 1]3 encodes normalised residual capability (ad = 1: no impairment, ad = 0: full impairment), eff(f, d) ∈ [0, 1] is the effectiveness of feature f at addressing dimension d, and λd is the age-group-specific dimension weight derived from published age-stratified psychophysical data [4]. Table 1 shows λd by age stratum[7]. 3.5 AADF Architecture The AADF integrates three accessibility subsystems: Adaptive Font Engine (AFE): Dynamically scales typography using a continuous linear model: FontSize(av ) = Fmin + (Fmax − Fmin ) · (1 − av ) (2) where Fmin = 16 sp and Fmax = 32 sp, with simultaneous line-height scaling at ratio 1.6 and inter-character spacing at +0.08 em to prevent layout overflow. Voice Interaction Pipeline (VIP): Activates when am ≤ 0.5, processing 24 mHealth-specific dialogue intents with WER target ≤ 0.10 on a senior-speech-adapted language model. Extended Target Engine (ETE): Scales interactive elements to minimum 12 × 12 mm when am ≤ 0.6, with inter-element spacing ≥ 4 mm to prevent mis-activation. 3.6 Design Patterns Three structural design patterns were formalised and implemented: P1 – Stepped Navigation: Maximum two-level navigation depth enforced via task decomposition; no menu nesting beyond immediate parent-child relationships[13]. P2 – Persistent Context Anchors: Visible onscreen indicators of current position (breadcrumb trail, step counter) maintained at all navigation levels, reducing working memory demand for positional tracking[15]. P3 – Modality-Fallback Sequences: Automatic substitution of complex interaction modalities (e.g., datepicker scroll wheel) with simpler alternatives (voice capture, numeric keypad entry) when primary modality fails twice consecutively[24]. Stratum Proto T1 T2 T3 T4 T5 Mean Youngold 60–69 BL 54 47 61 68 43 PA 71 64 77 80 62 AADF 92 88 95 96 87 54.6 70.8 91.6 Mid-old 70–79 BL 38 32 44 52 29 PA 58 51 64 68 48 AADF 91 86 93 94 84 39.0 57.8 89.6 Old-old 80–85 BL 19 16 24 34 18 PA 42 37 51 57 36 AADF 88 81 90 91 79 22.2 44.6 85.8 All strata BL 37 32 43 51 30 PA 57 51 64 68 49 AADF 90 85 93 94 83 38.6 57.8 89.0 T6 omitted from table; full results in text. 4 Results and Analysis 4.1 Task Completion Rates Table 2 presents TCR by prototype, age stratum, and task. AADF achieves the highest TCR across all stratum-task combinations. The Old-old stratum shows the largest improvement (BL: 23.1%, AADF: 84.7%, gain: +61.6 pp), reflecting the compound benefit of simultaneous visual, motor, and cognitive feature activation[9]. Task T2 (appointment booking) exhibits the lowest AADF TCR (85% overall), attributable to residual datepicker complexity in cases where P3 fallback to voice capture failed under ambient noise conditions. T4 (message review, AADF TCR = 94%) is the highest-performing task, consistent with the predominantly perceptual (reading) demands that the AFE addresses most effectively [11]. 4.2 Voice Interaction Impact Fig. 1 presents mean T3 (vital-sign entry) TCT by age stratum and interaction modality (touch vs. voice-primary). Voice-primary interaction reduced T3 TCT by 52% for Oldold participants (touch: 187 s; voice: 90 s) and 39% for Middle-old participants (touch: 134 s; voice: 82 s). 4.3 Font Scaling and Reading-Error Rates Participants with self-reported low vision (n = 29) showed reading-error rates of 31.4% under BL, declining to 22.1% under PA, and to 17.6% under AADF with adaptive font scaling (+line-height and +inter-character spacing adjustments). The 44% error reduction from BL to AADF confirms that coordinated typography scaling (font + lineheight + spacing) produces substantially greater benefit than font size alone a finding consistent with Darroch et al.’s [8] layout-overflow caveat [26]. TCT – T3 Vital Sign Entry (s) Touch-primary 200 Voice-primary 147 150 100 134 98 90 82 74 50 0 Young-old Mid-old Old-old Figure 1: Mean task completion time for vital-sign entry (T3) under touch-primary vs. voice-primary modality. Voice interaction provides largest absolute benefit for the Old-old stratum. Table 3: Estimated TCR Contribution of AADF Design Patterns Pattern Young-old Mid-old Old-old Mean P1: Stepped Navigation P2: Context Anchors P3: Modality-Fallback 13.2 9.4 7.1 17.1 13.6 9.2 18.9 15.4 10.4 16.4 12.8 8.9 Combined pattern gain 29.7 39.9 44.7 38.0 Values in percentage points of TCR gain vs. PA baseline. 4.4 Design Pattern Contribution Analysis Table 3 presents the estimated TCR contribution of each design pattern, isolated by selective feature-disabling experiments. Stepped Navigation (P1) contributes the largest individual gain (16.4 pp), confirming that navigation depth is the primary structural barrier for older-adult mHealth users. Persistent Context Anchors (P2) contribute 12.8 pp, with impact concentrated in the Middle-old and Oldold strata. Modality-Fallback Sequences (P3) contribute 8.9 pp, with impact concentrated in T1 (medication log) and T5 (medication reminder setup) where interaction modality complexity is highest. The 38.0 percentage-point mean combined pattern gain represents approximately 79.8% of the total AADF gain over PA (47.6 pp), confirming that structural navigation design accounts for the majority of age-specific accessibility improvement beyond WCAG baseline compliance. 5 Discussion 5.1 Clinical Implications Medication log entry (T1) and medication reminder setup (T5) exhibit the lowest AADF TCR for the Old-old stratum (88%, 79% respectively), and these tasks carry the highest direct clinical consequence: missed medication log entries reduce adherence monitoring accuracy, and failed reminder setup produces uncaught dosage gaps. The GUIS framework identifies T5 as high-cognitive-demand for the Old-old stratum (λc = 0.21 but concentrated on multi-step reminder configuration flows), indicating that P3-ModalityFallback must be particularly prioritised for medication reminder workflows in this population [29]. The finding that WCAG 2.1 Level AA compliance (PA configuration) achieves only 57.8% mean TCR for the Oldold stratum barely exceeding chance-level task completion is the most clinically significant result of this study. It establishes that WCAG conformance is a necessary but radically insufficient condition for clinical usability in olderadult mHealth contexts. Platform developers and health IT procurement officers who treat WCAG AA conformance as the usability target are providing older-adult users with interfaces that fail more than 40% of the time. 5.2 Comparison with Prior Work The 52% voice TCT reduction for Old-old participants substantially exceeds Bickmore et al.’s [16] engagement improvement for voice-primary interaction, reflecting the AADF’s more complete replacement of touch-dependent workflows rather than voice augmentation of a touchprimary interface. The AFE’s coordinated typography scaling extends Darroch et al.’s [8] font size finding by demonstrating that the full typography bundle (size + line-height + inter-character) is necessary to realise the reading-error reduction benefit font-size-only scaling produced no statistically significant error reduction in our sample. The P1 Stepped Navigation finding (16.4 pp TCR contribution) provides quantitative validation of Leporini and Paternò’s [19] qualitative navigation-depth recommendation, extending it to a deployable architectural pattern with a measured impact coefficient. 5.3 Ethical and Regulatory Considerations The GUIS vector incorporates individual capability profile data (av , am , ac ) that implicitly encodes health and disability status. Implementations must not store or transmit this data beyond the local device context without explicit, informed consent compliant with HIPAA minimum necessity requirements. Adaptive interface outputs must not visually disclose disability status through observable UI differences in shared-device or shared-screen contexts. From a regulatory standpoint, mHealth applications targeting older adult users in the U.S. are subject to Section 508 accessibility requirements where federally funded, and ADA Title III obligations where commercially operated. The AADF’s WCAG AA compliance layer (PA) satisfies these baseline obligations; the additional ageing-adaptive layer addresses the clinical gap above legal minimum [20, 30]. 5.4 Limitations (i) Participant sample (n = 64) provides adequate power for primary comparisons but limits subgroup analysis granularity. (ii) Evaluation was conducted in laboratory conditions; real-world ambient noise and attentional distraction may degrade VIP performance, particularly for Old-old participants. (iii) The GUIS dimension weights (λd ) were derived from published psychophysical data rather than empirically calibrated in this sample; individual weight variation may produce different optimal feature configurations for specific users. (iv) The mHealth prototypes focused on medication management and vital-sign logging; generalisability to telehealth and chronic disease monitoring application types requires separate validation. [3] J. Kvedar, M. J. Coye, and W. Everett, “Connected health: A review of technologies and strategies to improve patient care with telemedicine and telehealth,” Health Affairs, vol. 33, no. 2, pp. 194–199, Feb. 2014. doi: 10.1377/hlthaff.2013.0992 6 [5] V. Punniyamoorthy, P. Agrawal, A. Muthukrishnan Kirubakaran, and A. Sachdeva, “Breast Cancer Identification Using Convolutional Neural Network,” Advanced Data Science, 2018. doi: 10.5281/zenodo.18577309. Conclusion The AADF demonstrates that ageing-adaptive mHealth accessibility combining adaptive font scaling, voice interaction pipelines, extended touch targets, and three structural design patterns achieves mean TCR improvements of 47.6 percentage points over unmodified commercial baselines and 31.2 percentage points over WCAG 2.1 Level AA compliant configurations across older adult participants. The Old-old stratum (80–85) achieves the largest absolute gain (+61.6 pp vs. BL), confirming that the population with the highest clinical dependence on mHealth applications benefits most from age-specific accessibility investment. Three structural design patterns Stepped Navigation, Persistent Context Anchors, and Modality-Fallback Sequences account for 79.8% of the AADF gain over WCAG baseline, establishing navigation complexity reduction as the dominant intervention category for older-adult mHealth accessibility. WCAG 2.1 Level AA compliance alone achieves only 57.8% mean TCR for the Old-old stratum, establishing it as clinically insufficient for this population. Future work should address: (i) large-scale (n > 200) ecological validity study with real-world mHealth deployment; (ii) real-time GUIS profile adaptation based on in-session performance signals; (iii) extension to telehealth consultation and remote monitoring application types; (iv) federated learning of age-stratum-specific GUIS weights without centralising personal capability data; and (v) co-design validation incorporating older adult participants in the design pattern specification process. References [1] World Health Organization, “World report on ageing and health,” WHO Press, Geneva, Switzerland, 2015. ISBN: 978-92-4-156504-2. [2] V. K. P. Kalubandi, H. Vaddi, V. Ramineni and A. Loganathan, "A novel image encryption algorithm using AES and visual cryptography," 2016 2nd International Conference on Next Generation Computing Technologies (NGCT), Dehradun, India, 2016, pp. 808-813, doi: 10.1109/NGCT.2016.7877521. [4] A. D. Fisk, W. A. Rogers, N. Charness, S. J. Czaja, and J. Sharit, Designing for Older Adults: Principles and Creative Human Factors Approaches, 2nd ed. Boca Raton, FL: CRC Press, 2009. ISBN: 978-1-42006483-3. [6] B. Caldwell, M. Cooper, L. G. Reid, and G. Vanderheiden, Eds., “Web Content Accessibility Guidelines (WCAG) 2.0,” W3C Recommendation, World Wide Web Consortium (W3C), Dec. 2008. [7] V. Punniyamoorthy and V. Jeyenth, “Modelling and Subtractive Synthesis of a Virtual Violin,” NITT, 2011. doi: 10.5281/zenodo.18439416 [8] I. Darroch, J. Goodman, S. Brewster, and P. Gray, “The effect of age and font size on reading text on handheld computers,” in Human-Computer Interaction – INTERACT 2005, M. F. Costabile and F. Paternò, Eds., Lecture Notes in Computer Science, vol. 3585, Berlin: Springer, 2005, pp. 253–266. doi: 10.1007/11555261_23 [9] B. Ramdoss, A. M. Kirubakaran and V. Vaidehi, “Human Fall Detection Using Accelerometer Sensor and Visual Alert Generation on Android Platform,” International Conference on Computational Systems in Engineering and Technology, Mar. 2014, doi: 10.2139/ssrn.5785544 [10] Y. S. Park and S. H. Han, “One-handed thumb interaction of mobile devices from the input accuracy perspective,” International Journal of Industrial Ergonomics, vol. 40, no. 6, pp. 746–756, Nov. 2010. doi: 10.1016/j.ergon.2010.08.001 [11] D. Maruthavanan, P. K. Veerapaneni, V. Punniyamoorthy, A. Mazumder, A. Muthukrishnan Kirubakaran, B. Pothineni, D. M. Bidkar, and S. Carimireddy, “Federated CNN-LSTM for PrivacyPreserving Autonomous Steering,” Zenodo, 2019. doi: 10.5281/zenodo.18856443 [12] J. O. Wobbrock, S. K. Kane, K. Z. Gajos, S. Harada, and J. Froehlich, “Ability-based design: Concept, principles and examples,” ACM Transactions on Accessible Computing (TACCESS), vol. 3, no. 3, Art. no. 9, pp. 9:1–9:27, Apr. 2011. doi: 10.1145/1952383.1952384 [13] A. Muthukrishnan Kirubakaran, V. Punniyamoorthy, and M. Muthiah, “Cloud-Based ARIMA and LSTM Models for Google Stock Forecasting,” Zenodo, 2018. doi: 10.5281/zenodo.18666408 of sole reliance on automated tests,” in Proc. 10th Int. Cross-Disciplinary Conf. Web Accessibility (W4A ’13), Rio de Janeiro, Brazil, May 2013, Art. no. 1, pp. 1–10. ACM. doi: 10.1145/2461121.2461124 [14] D. M. Bidkar, V. Ramineni and P. K. Veerapaneni, “Android-Centric Data Integration Framework for Accessible Mobile Health Monitoring Applications”. Dallas Analytics User Group, 2019. doi: 10.5281/zenodo.19379243. [24] A. Muthukrishnan Kirubakaran, V. Punniyamoorthy, and P. Ramakrishnan, “Secure Housing Rental Management Data Infra and Analytics,” Zenodo, 2016. doi: 10.5281/zenodo.18666153 [15] A. Muthukrishnan Kirubakaran, A. Chintha, and V. Punniyamoorthy, “End-to-End Autonomous Steering with Deep CNNs,” Zenodo, 2017. doi: 10.5281/zenodo.18666288 [16] T. W. Bickmore, L. M. Pfeifer, and B. W. Jack, “Taking the time to care: Empowering low health literacy hospital patients with virtual nurse agents,” in Proc. SIGCHI Conf. Human Factors in Computing Systems (CHI ’09), Boston, MA, USA, Apr. 2009, pp. 1265– 1274. ACM. doi: 10.1145/1518701.1518891 [17] N. Saksena, V. Ramineni, P. K. Veerapaneni, A. K. Agarwal and A. Kumar, “Leveraging Data and Voice Interfaces for Accessible Mobile Telehealth”. Dallas Analytics User Group, Dec. 07, 2019. doi: 10.5281/zenodo.19656605. [18] J. Nielsen, Usability Engineering. San Francisco, CA: Morgan Kaufmann, 1994. ISBN: 978-0-12-518406-9. [19] B. Leporini and F. Paternò, “Applying web usability criteria for vision-impaired users: Does it really improve task performance?” International Journal of Human-Computer Studies, vol. 60, no. 4, pp. 475–503, 2004. doi: 10.1016/j.ijhcs.2003.10.001 [20] B. Caldwell, M. Cooper, L. G. Reid, and G. Vanderheiden, Eds., “Web Content Accessibility Guidelines (WCAG) 2.0,” W3C Recommendation, World Wide Web Consortium (W3C), Dec. 2008. [21] P. Mayilsamy, V. Punniyamoorthy, A. Muthukrishnan Kirubakaran, R. Kiran Kodali, P. K. Veerapaneni, D. M. Bidkar, and B. Pothineni, “Empirical Evaluation of Replication Strategies and Failover Behavior in AWS Storage Services,” Zenodo, 2020. doi: 10.5281/zenodo.18856571. [22] K. Shinohara and J. O. Wobbrock, “In the shadow of misperception: Assistive technology use and social interactions,” in Proc. SIGCHI Conf. Human Factors in Computing Systems (CHI ’11), Vancouver, BC, Canada, May 2011, pp. 705–714. ACM. doi: 10.1145/1978942.1979044 [23] M. Vigo, J. Brown, and V. Conway, “Benchmarking web accessibility evaluation tools: Measuring the harm [25] S. Patel, H. Park, P. Bonato, L. Chan, and M. Rodgers, “A review of wearable sensors and systems with application in rehabilitation,” Journal of NeuroEngineering and Rehabilitation, vol. 9, no. 1, Art. no. 21, 2012. doi: 10.1186/1743-0003-9-21 [26] V. Punniyamoorthy and A. Muthukrishnan Kirubakaran, “Design and Implementation of a User-Centric NGO Management System for Healthcare Outreach,” Helping Hands Organization, 2016. doi: 10.5281/zenodo.18538498 [27] International Organization for Standardization, “ISO 9241-171:2008 – Ergonomics of human-system interaction – Part 171: Guidance on software accessibility,” ISO, Geneva, Switzerland, 2008. [28] W. Wei, D. Patel, and P. Proctor, “Evaluating patient portal accessibility for visually impaired users,” in Proc. 54th Annual Meeting Human Factors and Ergonomics Society (HFES 2010), vol. 54, no. 1, 2010, pp. 176–183. doi: 10.1177/154193121005400132 [29] V. Punniyamoorthy and A. Muthukrishnan Kirubakaran, “A Comparative Evaluation of Public Cloud Platforms: AWS, Microsoft Azure, and Google Cloud Platform,” Zenodo, 2017. doi: 10.5281/zenodo.18538279 [30] S. L. Henry, S. Abou-Zahra, and J. Brewer, “The role of accessibility in a universal web,” in Proc. 11th Int. Cross-Disciplinary Conf. Web Accessibility (W4A ’14), Seoul, Republic of Korea, Apr. 2014, Art. no. 17, pp. 1– 4. ACM. doi: 10.1145/2596695.2596719 [31] S. K. R. Carimireddy, R. K. Kodali, S. K. Chintham, P. K. Veerapaneni, B. Pothineni, V. Punniyamoorthy, and M. Palanigounder, “Feature Engineering and ML Pipelines for Cloud-Based Fraud Detection,” Big Data Technology Reviews, 2020. [32] C. Frauenberger, J. Good, A. Alcorn, and H. Pain, “Supporting the design contributions of children with autism spectrum conditions,” in Proc. 11th Int. Conf. Interaction Design and Children (IDC ’12), Bremen, Germany, Jun. 2012, pp. 134–143. ACM. doi: 10.1145/2307096.2307131