📄 medicine

Modeling the Effects of Different Interventions on Kinesiophobia, Pain, and Quality of Life in Women with Lumbar Disc Herniation Using Artificial Neural Networks

This study demonstrates that swimming exercise significantly reduces pain compared to physical therapy or inactivity in women with lumbar disc herniation and validates the use of Artificial Neural Networks to accurately predict physical quality-of-life scores based on pain and kinesiophobia levels.

Original authors: Servet Özoruç, Ayşenur Gürgen, Dilek Sevimli

Published 2026-08-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Servet Özoruç, Ayşenur Gürgen, Dilek Sevimli

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine your body is a high-performance sports car. Usually, it's designed to run, jump, and twist. But sometimes, a part of the engine—specifically the cushioning between the gears in your lower back—gets squished and pops out of place. This is called a lumbar disc herniation, and it's like a pothole in the road that makes driving (or walking) incredibly painful. When this happens, many people get scared to move. They start thinking, "If I bend, it will break!" This fear is called kinesiophobia. It's like a passenger in the car screaming "Don't turn the wheel!" even when the road is clear. Scientists have long known that moving helps fix the engine, but they've been trying to figure out which kind of movement works best: is it the gentle, controlled repair of physical therapy, the fun, water-based glide of swimming, or is doing nothing at all (sitting on the couch) the real culprit? To solve this, researchers are using a special kind of computer brain called an Artificial Neural Network (ANN). Think of an ANN not as a robot, but as a super-smart detective that can spot hidden patterns in a messy pile of clues that a regular human might miss. It's like trying to guess the weather by looking at clouds, wind, and humidity all at once, rather than just checking the temperature.

In this study, a team of researchers decided to put this detective to work on 54 women who were dealing with this specific back problem. They split the women into three teams to see how different lifestyles affected their pain, their fear of moving, and their overall quality of life. One team was the "Swimmers," who had been swimming regularly for at least six weeks. The second team was the "Therapy Group," who were getting standard physical therapy treatments. The third team was the "Inactive Group," women who had the condition but didn't exercise at all. The researchers measured how much pain they felt, how afraid they were to move, and how well they could do daily tasks. Then, they fed all this data into their computer detective to see if it could predict how well a woman would feel based on her pain and fear levels.

The results were pretty clear, especially when it came to pain. The women who were swimming had the lowest pain scores. In fact, the computer analysis showed that the swimming group felt significantly less pain than the women who were inactive. The swimming group was also less painful than the physical therapy group, though that difference wasn't quite as statistically loud as the one against the inactive group. Interestingly, the study found that swimming didn't necessarily make the fear of moving disappear completely compared to the other groups, nor did it instantly skyrocket their quality-of-life scores to a perfect 100. The fear and the general "feeling good" scores were actually quite similar across all three groups. This suggests that while swimming is a fantastic way to turn down the volume on pain, fixing the fear and the overall quality of life might need a little more than just exercise alone.

But here is where the computer detective really shines. The researchers built three different models to see if they could predict a woman's physical quality of life just by knowing her pain level and her fear level. They tested these models on the different groups. The model comparing Swimmers and the Physical Therapy group was the star of the show. It was incredibly accurate, with a correlation score of 0.991 (which is almost a perfect match). This means the computer could look at how much pain and fear a swimmer had and guess their quality of life with almost zero error. The model for the Physical Therapy vs. Inactive group was also very strong. However, the model for Swimmers vs. Inactive was a bit messier, with slightly higher errors, suggesting that the gap between moving in water and doing nothing is a bit more complex to predict than the differences between the two active groups.

The big takeaway is that swimming seems to be a powerful tool for managing back pain in women with this condition, likely because the water takes the weight off the spine, letting them move without the "crunch" of gravity. The study suggests that while swimming is great for pain, it might not be a magic bullet for the fear of moving or the overall quality of life on its own; those might need extra help, like talking therapies or counseling. The study also proves that these fancy computer networks are really good at mapping out how pain and fear connect to how well a person feels, especially when comparing active treatments. The researchers are careful to say that because they only looked at 54 women, these computer models need to be tested on bigger groups to make sure they work for everyone. But for now, it looks like if you're stuck in pain and afraid to move, getting into the pool might be a very smart first step.

Technical Summary: Modeling the Effects of Different Interventions on Kinesiophobia, Pain, and Quality of Life in Women with Lumbar Disc Herniation Using Artificial Neural Networks

Problem Statement
Lumbar disc herniation (LDH) is a prevalent structural pathology causing pain, functional limitation, and reduced quality of life, particularly among women. The etiology involves complex interactions between physical factors (e.g., obesity, inactivity) and psychological factors, specifically kinesiophobia (fear of movement). While conservative treatments such as physical therapy and swimming are known to be beneficial, the relationships between intervention types, pain levels, kinesiophobia, and quality of life are often nonlinear and multidimensional. Conventional statistical methods may struggle to fully capture these complex patterns. This study addresses the need to evaluate the relative effects of swimming exercise, physical therapy, and an inactive lifestyle on these variables and to model the predictive relationship between pain/kinesiophobia and physical quality of life using Artificial Neural Networks (ANNs).

Methodology
The study employed a descriptive, comparative, and cross-sectional design involving 54 female volunteers diagnosed with LDH (mean age: 52.83 ± 13.30 years). Participants were divided into three purposively sampled groups (n=18 each):

Swimming Group: Women engaging in regular swimming exercises for at least six weeks.
Physical Therapy Group: Women receiving standard physical therapy programs at an outpatient clinic.
Inactive Group: Physically inactive women with LDH.

Data Collection:
Data were gathered using:

Visual Analogue Scale (VAS): To measure pain intensity.
Tampa Scale for Kinesiophobia (TSK): To assess fear of movement.
SF-12 Health Survey: To evaluate quality of life, specifically the Physical Component Summary (PCS) score.

Statistical and Modeling Analysis:

Group Comparisons: Normality was assessed via the Shapiro–Wilk test. Group differences were analyzed using MANOVA and the Kruskal–Wallis test, followed by Dunn–Bonferroni post-hoc tests.
Artificial Neural Network (ANN) Modeling: Separate ANN models were developed for each pairwise group comparison (Swimming vs. Physical Therapy, Physical Therapy vs. Inactive, Swimming vs. Inactive).
- Inputs: Total Kinesiophobia Score and Pain Score.
- Output: SF-12 Physical Health Score.
- Architecture: The Levenberg–Marquardt algorithm was used for training. Data were split into training (80%), validation (10%), and testing (10%) sets. Network architectures were optimized by varying hidden neurons (1–20) over 1,000 independent runs.
- Performance Metrics: Models were evaluated using Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), and the correlation coefficient (R).

Key Results

Demographics: No significant differences were found among the groups regarding age, height, weight, BMI, or disease duration, ensuring a valid basis for clinical comparison.
Clinical Outcomes:
- Pain: A statistically significant difference was observed among groups ( $p = 0.002$ ). The inactive group exhibited the highest pain levels (mean rank 35.61), while the swimming group had the lowest (mean rank 17.75). Post-hoc analysis confirmed a significant difference between the swimming and inactive groups ( $p = 0.002$ ), but not between swimming and physical therapy ( $p = 0.085$ ) or physical therapy and inactive ( $p = 0.638$ ).
- Kinesiophobia and Quality of Life: No statistically significant differences were found among the three groups for total kinesiophobia scores ( $p = 0.059$ ) or physical quality-of-life scores ( $p = 0.130$ ).
ANN Modeling Performance:
- Swimming–Physical Therapy Model: Optimal with 8 hidden neurons. Achieved MSE = 0.675, MAPE = 3.242%, and R = 0.991.
- Physical Therapy–Inactive Model: Optimal with 11 hidden neurons. Achieved MSE = 0.705, MAPE = 3.410%, and R = 0.991.
- Swimming–Inactive Model: Optimal with 7 hidden neurons. Achieved MSE = 1.429, MAPE = 5.363%, and R = 0.973.
- All models demonstrated high agreement between observed and predicted physical health scores, with the Swimming–Physical Therapy model showing the highest predictive accuracy.

Significance and Claims
The paper claims that swimming exercise may serve as a viable option for pain management in women with lumbar disc herniation, demonstrating lower pain scores compared to an inactive lifestyle and comparable results to physical therapy. Furthermore, the study establishes that ANNs can effectively model the nonlinear relationships between pain, kinesiophobia, and physical quality of life, predicting SF-12 physical health scores with high accuracy (R > 0.97).

The authors emphasize that while physical interventions reduce pain, they do not necessarily immediately resolve kinesiophobia or quality-of-life deficits, suggesting a need for multidisciplinary approaches that include psychological and educational components. The study concludes that while the ANN approach successfully identified group-specific data patterns, the findings are limited by the sample size and should be validated in larger, more diverse populations before being generalized for clinical planning.

Technical Summary: Modeling the Effects of Different Interventions on Kinesiophobia, Pain, and Quality of Life in Women with Lumbar Disc Herniation Using Artificial Neural Networks

More like this