## GH-METHODS

Math-Physical Medicine

### NO. 333

Comparison of cardiovascular disease or stroke risk probability percentages using annualized Excel model and daily APP model via GH-Method: math-physical medicine

Corresponding Author: Gerald C. Hsu, eclaireMD Foundation, USA.

Abstract
The author uses his developed GH-Method: math-physical medicine (MPM) approach to conduct detailed analysis and numerical calculation of his risk probability percentages (Risk %) of having cardiovascular diseases (CVD) or stroke via two different models or devices.

In early 2018, he started the development of his CVD/Stroke risk prediction model utilizing big data collected from his customized software for chronic diseases.  He then developed a mathematical model by using Microsoft Excel on a PC to further process the collected data in order to obtain the CVD/Stroke risk percentages.  After three years of utilizing the PC-based Excel model, he decided in mid-2020 to perform a software programming task of converting his existed mathematical model, procedures, and equations into an iPhone-based APP model.

In this article, he compares the end results of these two sets of calculated CVD/Stroke risk probability percentages via two different models on separate devices.  The purpose is twofold: to see how big the numerical differences are between the different models, and if the differences are significant, then further investigation behind the numerical deviations is required.  If the numerical deviations are small, then he will stop his investigation and use the iPhone APP model.

This article describes the numerical deviations of CVD/Stroke risk percentage using two different models and devices, Excel model on a PC vs. APP model on an iPhone.  The final risk percentage results have shown an insignificant deviation of ~1%.  This small deviation is a result from inherited data incompleteness and differences on data collection methods; therefore, the developed APP model can be utilized by patients with chronic diseases to guesstimate their risk probability percentage of having cardiovascular disease or stroke.

Introduction
The author uses his developed GH-Method: math-physical medicine (MPM) approach to conduct detailed analysis and numerical calculation of his risk probability percentages (Risk %) of having cardiovascular diseases (CVD) or stroke via two different models or devices.

In early 2018, he started the development of his CVD/Stroke risk prediction model utilizing big data collected from his customized software for chronic diseases.  He then developed a mathematical model by using Microsoft Excel on a PC to further process the collected data in order to obtain the CVD/Stroke risk percentages.  After three years of utilizing this PC-based Excel model, he decided in mid-2020 to perform a software programming task of converting his existed mathematical model, procedures, and equations into an iPhone-based APP model.

In this article, he compares the end results of these two sets of calculated CVD/Stroke risk probability percentages via two different models on separate devices.  The purpose is twofold: to see how big the numerical differences are between the different models, and if the differences are significant, then further investigation behind the numerical deviations is required.  If the numerical deviations are small, then he will stop his investigation and use the iPhone APP model.

Methods
1. Background
To learn more about the GH-method: Math-physical medicine method (MPM) developed by the author, readers can review his article, Biomedical research methodology based on GH-Method: math-physical medicine (No. 310).  In addition, the outlined history of his chronic diseases, diabetes research, and application tools development can be found in another article, Glucose trend pattern analysis and progressive behavior modification of a T2D patient using GH-Method: math-physical medicine (No. 305).

The author is a 73-year-old male who has a history of three severe chronic diseases for 25 years.  In addition, he experienced five cardiac episodes during 1994 – 2008 and was diagnosed with an acute renal problem in early 2010.  He also suffered from foot ulcer, bladder infection, diabetic retinopathy, and hypothyroidism.  He weighed 220 lbs. in 2000 and his HbA1C level was 10.0% in 2010.

In 2014, he developed a complex  mathematical model of metabolism and also started a stringent lifestyle management program.  As a result, his overall health conditions have been noticeably improving since 2015.  Around 2013, he started to reduce the dosage of his three prescribed diabetes medications.  By 12/8/2015, he completely stopped taking them.  During the entire period of 2016-2019, his HbA1C average value was 6.6% without medication. During the recent COVID-19 quarantine period from 1/19/2020 to 9/26/2020, his HbA1C has further decreased to 6.1%.

2. Metabolism Model
In 2014, he applied topology concept of mathematics and finite-element method of engineering, to develop a ten-dimensional mathematical model of metabolism which contains four output categories such as weight, glucose, BP, lipids and other lab-tested data, along with six input categories including food, water intake, exercise, sleep, stress, and daily life routines.  Furthermore, these ten categories contain about 500 detailed elements.  He has calculated individual score for each  metabolism category with a symbol of mi where I = 1 through 10.

Finally, he defined a new parameter, metabolism index (MI), as the combined score of the above 10 metabolism categories.  Since 2012, he has collected around 2 million data of his own biomedical conditions and personal lifestyle details.  However, he only started to collect his complete lifestyle data since 5/1/2015 after he has completed his development of metabolism model and post-prandial plasma glucose (PPG) in 2014-2015.

3. CVD/Stroke Risk Model
In 2017, he developed a few suitable algorithms containing some different weighting factors which include a patient’s baseline data such as gender, age, race, family genetic history, medical history, bad habits (cigarette smoking, alcohol drinking, illicit drugs), BMI (for obesity, m1), and waistline.  After continuously collecting sufficient input data for nine years, he then conducted the following three sets of calculations regarding risk of having a CVD or Stroke.

The first set consists of the medical conditions approach – individual category scores for diabetes (m2), hypertension (m3), hyperlipidemia and others (m4).  These three metabolic disorder values contain the patient’s biomedical data and hospital laboratory test results.  Through his previous research work since 2015, he already detected that glucose is the “principal criminal” and blood pressure with lipids as the “accessory criminals” in terms of metabolic disorders induced chronic disease complications, specifically CVD, stroke, renal problems (CKD), diabetic retinopathy, foot ulcer, bladder infection, hypothyroidism, and some cancers.  More precisely, his mathematical model for CVD or stroke includes two scenarios related to arteries.  The first scenario is the artery blockage situation which involves diabetes (glucose), hypertension (blood pressure or BP), and hyperlipidemia (lipids), where he applied his acquired solid mechanics and fluid dynamics concepts.  The second scenario is the artery rupture situation which involves diabetes (glucose), and hypertension (BP), where he applied his acquired nonlinear solid dynamics and fracture mechanics concepts. With regard to mathematical model related to micro blood vessels, it would utilize a different approach to proceed with the investigation.

The second set involves lifestyle details approach – individual m5 through m10 which are the fundamental causes of many diseases, especially the four medical conditions of m1 through m4, directly or indirectly.  In this approach, he includes the following three sub-groups with a total of nine detailed sub-categories of lifestyle details.

• (B-1) Three food values: quantity (meal portion), quality (nutritional balance), and carbs/sugar intake amount; for this B-1, the lesser amount would yield a better score.
• (B-2) Two exercises and water intake: daily walking steps for general heath and post-meal waking steps for diabetes control, drinking water intake in ml or cc; for this B-2, the more amount would yield a better score.
• (B-3) Three other values: sleep, stress, and daily life routines for general health of body and organs; for this B-3, the lower score would yield a better result.

The third set includes the combined MI approach – MI is a combined score of m1 through m10 using concept of mathematical topology and finite element approximation method from both structural and mechanical engineering.

With his developed mathematical risk assessment and probability tool, he can obtain three separate risk probability percentages associated with each of these three approaches, MI, medical conditions, and lifestyle details.  As a result, this tool offers a range of the risk probability predictions of having CVD or stroke, depending on the patients’ medical conditions, lifestyle details, and/or the combined metabolism impact on our body.

His risk assessment is based on his medical conditions data collected since 1/1/2012, while his risk assessment is based on his lifestyle details data collected from 5/1/2015.  All of these collected data prior to 9/26/2020 are used in this analysis project.  It should be pointed out that his lifestyle details from 2015 include 8 months of data, and his medical conditions and lifestyle details from 2020 include 9 months of data.  For source data reliability, he decided not to utilize all of his partially collected data from 2012 through 2014;  therefore, this article only presents his input data and output results during the time span of 1/1/2015 through 9/26/2020.

For the past two years, the author published approximately 17 medical papers regarding risk of having a CVD or stroke using annualized average data and Excel calculations (the Excel Model).  Recently, he decided to take the model mentioned from above, procedures and equations and convert them into a software with a format of iPhone APP (the APP Model).  He made this CVD/Stroke APP link directly with his existing chronic disease analysis software on his iPhone (the Chronic APP).  In this way, he could use his complete daily dataset as direct inputs into his CVD/Stroke APP to calculate the risk probability percentages.

Unfortunately, there were some empty data in his database due to small amount of missing data entry or data correction due to data error resulting from the glucose measuring device.  Therefore, it is inevitable that some degree of numerical deviations between these two different models, the Excel model and the APP model.  If the numerical deviations are significant, he will then attempt to address these differences and their degree of significance from different viewpoints of input data checking, biomedical disease interpretation, and lifestyle details investigation.

Results
In Figure 1, it shows the background data table of his analysis results. The Excel results vs. APP results are placed side by side for the MI, medical conditions, and lifestyle details scores.

###### Figure 1: Background data table of CVD/stroke risk probability % (2015-2020)

Figures 2 and 3 reveal the CVD/Stroke risk percentage based on MI, medical conditions, and lifestyle details via the APP model and Excel model, respectively.

###### Figure 3: Comparison among MI-based, medical-based, and lifestyle-based using APP model (2015-2020)

Figures 4, 5, and 6 reflect the CVD/Stroke risk percentage using both the APP model and Excel model, based on MI, medical conditions, and lifestyle details, respectively.

###### Figure 6: Comparison between APP and Excel of lifestyle-based CVD/Stroke risk % (2015-2020)

Here are some summarized observations from these five figures.

• The general trend of the three bar sets are extremely similar, i.e. with similar wave patterns (Figures 2 and 3). The risk bar starts from a higher risk in 2015, reaching a lower risk in 2017, then rises again to higher risks in 2018 and 2019 due to hectic travel schedules, and finally dropping to the lowest risk level in 2020 due to COVID-19 quarantined life.
• The comparison between the Excel model and the APP model, their 6-year averages among MI, medical, lifestyle are within 1%, which is insignificant. It should be mentioned that lifestyle contains around 85% of his total ~2 million metabolism data with medical containing the remaining 15%.
• From Figures 4, 5, and 6, the comparisons between the APP model and Excel model are based on three approaches, MI, medical, and lifestyle, where they have shown similar wave patterns (curve characters), and insignificant deviation of 1%. This insignificant difference is even more visible in Figures 3, 4, and 5 compared to Figures 2 and 3.
• The MI-based prediction offers the “best” CVD/Stroke risk assessment. However, either medical-based or lifestyle-based assessments can still provide some different angles to investigate and obtain useful information regarding the subject of CVD/Stroke risk percentage.
• According to the author’s acquired knowledge and research experience on this subject, patients would face a more serious risk of having a CVD or stroke if their risk percentage reaches to 70% or 75%. The author has had five cardiac episodes in the past when his guesstimated CVD risk percentage reached to a range between 80% and 100% during 2000 through 2014.  (These percentages are calculated using his specially developed model and equations; therefore, please do not use the common sense for the numerical percentage to make an incorrect judgement).

Conclusion
This article describes the numerical deviations of CVD/Stroke risk percentage using two different models and devices, Excel model on a PC vs. APP model on an iPhone.  The final risk percentage results have shown an insignificant deviation of ~1%.  This small deviation is a result from inherited data incompleteness and differences on data collection methods; therefore, the developed APP model can be utilized by patients with chronic diseases to guesstimate their risk probability percentage of having cardiovascular disease or stroke.

References

1. Hsu, Gerald C., eclaireMD Foundation, USA: “Biomedical research methodology based on GH-Method: math-physical medicine (No. 310)”
2. Hsu, Gerald C., eclaireMD Foundation, USA: “Glucose trend pattern analysis and progressive behavior modification of a T2D patient using GH-Method: math-physical medicine (No. 305)”
3. Hsu, Gerald C., eclaireMD Foundation, USA.  255: “Using mathematical model of metabolism to estimate the risk probability of having a cardiovascular diseases or stroke during 2010-2019  (GH-Method: math-physical medicine)
4. Hsu, Gerald C., eclaireMD Foundation, USA. No.314: “Detailed lifestyle contributions on risk probability % of having a CVD or stroke using GH-Method: math-physical medicine”
5. Hsu, Gerald C., eclaireMD Foundation, USA.   43 “Using GH-Method: math-physical medicine to investigate the triangular dual-correlations among weight, glucose, blood pressure with a Comparison of 2 Clinic Cases”
6. Hsu, Gerald C., eclaireMD Foundation, USA.  13: “Using GH-Method: math-physical medicine and signal processing techniques to predict PPG”
7. Hsu, Gerald C., eclaireMD Foundation, USA.  289: “Risk Probability of Atherosclerosis, Cardiovascular Disease, and Stroke during the COVID-19 period using GH-Method: math-physical medicine”
8. Hsu, Gerald C., eclaireMD Foundation, USA.  No. 273: “Using glucose and its associated energy to study the risk probability percentage of having a stroke or cardiovascular diseases from 2018 through 2020 (GH-Method: math-physical medicine)”
9. Hsu, Gerald C., eclaireMD Foundation, USA.   259: “Relationship between metabolism and risk of cardiovascular disease and stroke, risk of chronic kidney disease, and probability of pancreatic beta cells self-recovery using GH-Method: Math-Physical Medicine”