Look closer.

How much does a FinTech breach actually cost, and who is hit hardest?

An analysis of 47 incidents in 14 countries between 2008 and 2025, representing US$6.4 billion in reported losses, examines breach patterns, vulnerability factors and financial impact.

Method

Python, statistical analysis, structured datasets, literature synthesis, quantitative and text analysis.

Key findings

Third-party cybersecurity risk is rising. Crypto/DeFi incidents have ~11× higher median losses. The research develops a 5-stage cybersecurity risk assessment framework.

Limitations & context

Incident-level data is aggregated from public reports, and reported losses are not a complete measure of harm.

SSRN preprint

Earlier work: the published IJISRT study below laid the groundwork for this analysis.

What can major breaches teach us about protecting financial systems?

The earlier work behind the mixed-methods FinTech study, examining Capital One, Bangladesh Bank and Travelex.

Method

Case-study analysis and synthesis of 12 academic and industry sources; evaluation of encryption, blockchain and AI/ML threat detection.

Key findings

A comprehensive analysis of data breaches and approaches to cybersecurity in FinTech.

Limitations & context

Case-study findings provide context, rather than universal estimates of risk.

Vol. 10 Issue 4 · doi:10.38124/ijisrt/25apr1107

What changes when an intervention adapts to the child?

Secondary quantitative and thematic analysis of 20 published studies explores personalisation, engagement, performance and accessibility, and proposes an adaptive GroovAbility.

Method

A structured dataset; Python, TF-IDF, k-means clustering and PCA.

Key findings

Favourable outcomes in 93% of adaptive/personalised studies versus 20% of fixed-design studies. A 4-layer framework: Instrumentation · Feature definition · Adaptation logic · Evaluation.

Limitations & context

Publication bias, small sample sizes and methodological limits. These are literature-level comparisons, not a controlled GroovAbility trial.

GroovAbility research

Can a visual-guided routine make movement more accessible?

A 3-week pilot with 42 neurodiverse children across 2 special education centres measured participation, motor coordination and usability through educator observations and feedback.

Method

Short-term pilot testing with educator observations and feedback.

Key findings

Educator-observed improvement in balance 52.4%, coordination 57.1% and confidence 59.5%. Dance-based exercises were especially engaging.

Limitations & context

Short-term, observational testing; the findings do not establish long-term effects or causality.

GroovAbility pilot study

How do government, academics and journalists frame the same policy differently?

A computational text analysis of 55 documents across 7 source categories and 15 stakeholder categories compares how different sources describe the Swachh Bharat Mission, identifying 5 thematic clusters and contrasting sentiment across source types.

Method

Python preprocessing, VADER sentiment, TF-IDF and LDA topic modelling.

Key findings

Government, academic and journalistic sources frame sanitation policy in contrasting ways; automated sentiment struggles to capture complex social issues.

Limitations & context

A sentiment score is not a substitute for reading the context, and corpus selection shapes what the framing analysis can see.

CREST Gold Award · October 2026

When a district is declared open defecation free, does the data agree?

A follow-up comparing official open-defecation-free declarations with NFHS-5 sanitation data across all 704 Indian districts, with first-difference regression on child-health outcomes.

Method

District-level data assembly and first-difference regression analysis.

Key findings

42% of districts fall below 70% sanitation coverage — a measurable gap between declaration and outcome.

Limitations & context

In progress: I am rewriting the analysis code myself and developing it into a standalone paper.

Independent follow-up