Annals of Allied Health Sciences

The Role of Log-Transformation in Data Science and Statistical Analysis

2026-05-07T05:32:27+00:00

The importance and analytical power of log-transformation in data science lies in its ability to address several common data challenges, thereby, improving model accuracy and enriching analytical insights.¹ Log-transformation is a widely used technique in various fields (including statistics, economics, biology and environmental science) to address issues, such as skewed distribution, heteroscedasticity and multiplicative relationships between variables.^{2, 3} Hereunder are some key reasons why log-transformation is important in data science.

The majority of real-world datasets are typically skewed or biased, with most of the observations falling at one end of the scale. Skewed data can breach statistical tests and model assumptions, leading to biased estimates and inaccurate inferences.⁴ The log-transformation is, arguably, the most popular among the range of transformation options applied to convert skewed data into approximate normality.¹ However, log-transformation is only warranted if the original data follows a log-normal distribution or approximately so. If not, other transformation alternatives must be explored to meet the normality assumption.¹ To normalize skewed distributions and stabilize variance, variables with non-normal distributions can be log-transformed using the natural logarithm:

X_{log =}ln(X)

Regression analyses then can be performed using the following general model:

Y = β₀ + β₁ X_log + ϵ

Where Y represents the outcome variable, β₀ is the intercept, β₁ is the regression coefficient, and ϵ is the error term.

Variance stabilization and homogenization of error variance: In regression analyses, analysis of variance (ANOVA) and other parametric tests, an important assumption is homoscedasticity, meaning that the variance of errors remains constant across groups or values of independent variables. Violation of this assumption is known as heteroscedasticity, can produce biased standard errors, inaccurate p-values, and unreliable statistical inferences. ^{2, 5} The log-transformation of dependent variable or independent variable, or both is commonly used to stabilize variance and reduce heteroscedasticity, thereby improving the validity of inferential procedures.^{1, 6} However, transformation may not always fully resolve the problem; therefore, researchers should assess residual variance after transformation and consider alternative approaches when heteroscedasticity persists.⁷

Linearization of relationships: Log-transformation can linearize nonlinear relationships between variables, making them easier to interpret and model.⁸ In many cases, the relationship between variables may be multiplicative rather than additive. Taking the logarithm of one or both variables can transform multiplicative relationships into additive ones, enabling the use of linear regression models to estimate the effects of predictors on the outcome variable.⁹

Outlier Reduction: Extreme values or outliers have the potential to disproportionately influence statistical analyses and distort parameter estimates.¹⁰ Log-transformations reduce the impact of outliers by compressing or shrinking large values and spreading out small values, which improve the robustness of statistical tests and reduce the influence of extreme observations on the results.³

Interpretation of transformed coefficients: Log-transformation is particularly useful when interpreting changes in the variables with exponential growth or decline, such as population, income or biomarker levels, furthermore, log-transformation facilitates the interpretation of percentage changes.¹¹ When the outcome variable is log-transformed in a regression model, regression coefficients correspond to proportional (or percentage) changes in the original scale of outcome when exponentiated.¹¹ This approach makes it easier to interpret the effects of predictors in regression models and to compare relative changes across different variables, groups, or time periods, even if their original scales differ widely. Despite the fact that log transformation is an important element still it has several important limitations. First, it cannot directly handle zero or negative values, as the logarithm is undefined for these cases. Researchers must often add a constant before transformation, which can introduce arbitrariness and bias.^{1, 3} Second, the interpretation of transformed variables, as regression coefficients and statistical estimates become less intuitive because the coefficients represent multiplicative rather than additive effects and may complicate clinical or public health interpretation.⁷ Additionally, the log-transformation may not always correct the heteroscedasticity or normalize the data, especially in situations when the distributions contain extreme outlier values.³ However in some situations, the transformed data may still go against the presumptions needed for parametric analyses. Therefore, it is strongly advised that researchers should thoroughly evaluate the distribution of data and model assumptions. Furthermore, when necessary, consider the alternative statistical techniques.

In summary, log-transformation is a valid and useful tool to fix the common problems in data or relevant issues of uneven distributions, extreme values or outliers, heteroscedasticity, and nonlinear relationships. Furthermore, it can improve how well the statistical models work and make the results easier to understand. By changing the data into a form that fits close to the assumptions of statistical models, log-transformation may help researchers to make more accurate and meaningful conclusions. Therefore, it is always recommended to verify the assumptions following data transformation. To make the results understandable, practical and significant, it is also advised that researchers should try to bring the findings back to the original scale (before the log change) so that the results can be meaningful and useful in real life.

Prevalence of Impostor Syndrome and its association with anxiety among undergraduate students: a cross-sectional study

2026-04-28T03:14:15+00:00

Background: A psychological phenomenon known as Impostor Syndrome has recently attracted more attention in scholarly and scientific circles, especially due to its effects on medical students. This study was carried to determine the prevalence of impostor syndrome among undergraduate students at Rehman Medical Institute (RMI) and its association with anxiety among students of RMI.

Material and methods: An analytical cross-sectional study was conducted in RMI Allied Health Sciences, concentrating on the prevalence of Impostor Syndrome and its association with anxiety. Questionnaires were distributed in hard form across RMI disciplines (RCRS, RCN, and RCAHS) with a sample size of 302 in which all willing undergraduate first to final year students of Allied Health Sciences were included. Students who had anxiety and uses medication for anxiety or having panic attacks were excluded from our study. Convenience sampling was used for data collection. SPSS version 2022 was used for the analysis of the data.

Results: Data was gathered and analysed from a total of 302 participants. The mean score for anxiety was 10.12 ± 5.23 and for imposter syndrome was 62.04 ± 11.06. Chi-square analysis revealed a strong association between anxiety and impostor syndrome. The results of the analysis showed that there were no statistically significant differences in anxiety and Impostor Syndrome according to age, gender, discipline, or academic year.

Conclusion: The study's findings indicate that Impostor Syndrome was present in most undergraduate students to varying degrees and that anxiety and Impostor Syndrome were significantly correlated. The research found no significant differences in Impostor Syndrome levels by gender, age, or study year. The finding of a strong positive correlation between Impostor Syndrome and anxiety further substantiated the link between students' heightened anxiety levels and feelings of inadequacy.

Patients’ perspectives on physiotherapist’s empathy and its association with therapeutic alliance: a cross-sectional study

2026-07-01T05:57:14+00:00

Background: Empathy and therapeutic alliance play a central role in patient-centered physiotherapy and contribute to better treatment outcomes. However, evidence on the relationship between them remains limited, particularly in Pakistan. Therefore, this study aimed to examine the relationship between empathy and therapeutic alliance among patients receiving physiotherapy.

Material and methods: Convenience sampling was used to identify 373 participants for this cross-sectional study, which took place in the Physical Therapy Department of Rehman Medical Institute between June and November of 2024. The 10-item CARE questionnaire was used to measure empathy with a Cronbach’s alpha value of 0.95, while the 12-item Working Alliance Inventory–Short Revised (Cronbach’s alpha = 0.85) was used to measure therapeutic alliance. IBM SPSS Statistics v22 was used to analyze the data.

Results: Of the 373 participants, the majority were female (53.4%) and fell within the 31–40 age range (28.2%). 40.2% of patients had chronic diseases, whereas 79.1% of patients were receiving treatment for musculoskeletal disorders. The findings showed that a better therapeutic alliance between patients and therapists was substantially correlated with higher empathy levels. Pearson's r = 0.506, 95% CI [0.42, 0.58].

Conclusion: Higher therapist empathy is associated with a better patient-therapist interaction, as evidenced by a moderate positive correlation between empathy and TA. This implies that more effective, person-centered physiotherapy care can be supported by empathic communication, which can also increase trust and patient participation.

Challenges experienced by the parents of children with thalassemia at Liaquat University hospital, Hyderabad

2026-06-18T07:17:04+00:00

Background: Thalassemia an inherited genetic blood disorder that requires ongoing transfusion therapy and creates major psychological, social, economic, informational, and healthcare access challenges for caregivers. This study aimed to assess the challenges experienced by parents of children with thalassemia attending a tertiary care hospital in Hyderabad.

Material and methods: A descriptive cross-sectional study took place in the thalassemia ward of Liaquat University Hospital, Hyderabad, from Feb to July 2025. Convenience sampling was used to recruit 112 parents/guardians. Data were gathered using a structured questionnaire, and SPSS version 26 was used for data analysis. Chi-square tests with p < 0.05 were utilized, along with descriptive statistics (frequencies, percentages, means, and standard deviations).

Results: The study comprised 112 participants with almost equal males (50.9%) and females (49.1%). The highest age group was 31–40 years (33.9%); Psychological distress was common, including fatigue and restlessness (63.4%); uncertainty about the future (61.6%); and hopelessness (58.9%). 65.2% of respondents reported financial burden, 57.1% reported informational difficulties, 53.6% reported limited social support, and 56.3% reported healthcare access issues. There were significant associations between caregiver challenges and gender, education, residence, family type, and income (p < 0.05), but not with age.

Conclusion: Parents of children with thalassemia experience significant multidimensional burdens, with significant socioeconomic contributions. There is a need for specific interventions such as caregiver education, community or workplace psychosocial support, and better availability of health services to minimize caregiver burden.

Frequency of computer vision syndrome and its association with forward head posture among physical therapy; an analytical cross-sectional study

2026-05-28T13:16:58+00:00

Background: The rapid spread of digital technologies in the modern world has changed education, communication, and healthcare, making screens a necessary part of daily life. The increasing incidence of Computer Vision Syndrome (CVS) is a significant concern that negatively impacts the health of numerous people especially university students who engage in prolonged screen time for both academic and recreational purposes and therefore requires urgent attention. This study was designed to find out the frequency of computer vision syndrome and its association with forward head posture.

Material and methods: An analytical cross-sectional study comprising 153 physical therapy students of RCRS comprising a population of 251 students was conducted at RCRS. The sampling technique used was probability, i.e. simple random sampling. The CVS smart questionnaire and a goniometer were used as tools to find the frequency of CVS and FHP. With permission obtained priorly and data collected, SPSS was used for data analysis.

Results: A total of 84 participants out of 153 (84%) had CVS. 123 (80%) participants had forward head posture. A significant association was established between CVS and CVA (P value 0.000), CVS and screen usage duration (P value 0.003) as well as CVS and Semester (P value 0.039).

Conclusion: The frequency of CVS and FHP was found among the physical therapy students. A significant association was established between increased duration of screen usage and CVS.

Comparative effects of modified constraint induced movement therapy with motor relearning program on balance among stroke patients

2026-05-28T14:19:54+00:00

Background: Stroke often leads to permanent balance problems and are major causes of death and disability across the globe. Techniques such as physiotherapy involving motor relearning program (MRP) and modified constraint-induced movement therapy (MCIMT) help in the recovery of motor functions after a stroke. MCIMT involves repetitive movements using the affected limb in order to increase neuroplasticity, whereas MRP helps in relearning motor functions. This study was carried out to compare the effects of motor relearning program and modified constraint induced movement therapy on balance among stroke patients.

Material and methods: This quasi-experimental study was conducted at Care Physiotherapy Clinic, Lahore, with 20 purposively selected participants. Inclusion criteria involved ischemic stroke patients (≥5 months), aged 35–55, cognitively stable, able to stand independently, and receiving regular therapy. Participants were divided into two groups (mCIMT and MRP), and balance outcomes were assessed using the Berg Balance Scale. Interventions were conducted 3 days a week for eight weeks, with outcomes measured at baseline, 4 weeks, and 8 weeks.

Results: Both groups exhibited significant improvements in BBS scores over time. However, MCIMT group showed greater improvement in balance (BBS=p<0.05) compared to MRP group at the end of the study.

Conclusion: This study concluded that there was significant difference between both groups but the group receiving modified constraint induced movement therapy showed more pronounced improvement in balance compared to patients receiving motor relearning program.