CC BY
19
Список литературы на английском языке /References in English
1. Umergalin T.G., Iskakova Z.M. Komp'jutemoe modelirovanie i optimizacija proizvodstvennyh tehnologicheskih ustanovok [Computer modeling and optimization of production process units] // Izvestija JuFU. Tehnicheskie nauki [Izvestia SFU. Technical Sciences], 2005. № 1 (45) P. 43-44 [in Russian].
2. Akhmetov A.F. Razrabotka i issledovanie kombinirovannogo processa riforminga benzinovyh frakcij: dis.kand. tehn. nauk. [Development and investigation of the combined process of reforming gasoline fractions: dis. Tech. Sciences]. Ufa, 1975. 156 p. [in Russian].
3. Matkulova L.F., Musina A.R., Osipova A.G. Modelirovanie i optimizacija himiko-tehnologicheskih ustanovok [Modeling and optimization of chemical plants] // Teorija i praktika massoobmennyh processov himicheskoj tehnologii (Marushkinskie chtenija) sbornik nauchnyh trudov po materialam V Mezhdunarodnoj nauchnoj konferencii [Theory and practice of mass-exchange processes of chemical technology (Marushkin Readings) collection of scientific papers on the materials of the V International Scientific Conference]. Ufa, 2016. P. 156 [in Russian].
4. Umergalin T.G. Matematicheskoe modelirovanie osnovnyh himiko-tehnologicheskih processov: Uchebnoe posobie [Mathematical modeling of basic chemical-technological processes: Textbook]. Ufa: Izd-vo UGNTU, 2001. 61 p. [in Russian].
5. Umergalin T.G., Umergalina T.V. Raschet teplo-massoobmena stupeni kontakta mnogokomponentnoj smesi [Calculation of heat-mass transfer of a step of contact of a multicomponent mixture] // Bashkirskij himicheskij zhurnal [Bashkirsky chemical journal], 2016. Vol. 2. № 2. P. 41-43 [in Russian].
6. Umergalin T.G. Process sovmeshhennoj mnogostupenchatoj kondensacii i isparenija smesi [The process of combined multi-stage condensation and evaporation of the mixture]. Ufa: Bashkirskoe knizhnoe izdatel'stvo [Bashkir Publishing House], 1991. 150 p. [in Russian].
LONG-RUNNING PREDICTIONS IN PULSE MEASURING TASKS Abzalova L.R. (Russian Federation) Email: Abzalova426@scientifictext.ru
Abzalova Liliya Radikovna - Bachelor, DEPARTMENT COMPUTER SYSTEMS AND NETWORKS, HIGHER SCHOOL OF ECONOMICS, MOSCOW
Abstract: this paper outlines the way in which the aggregated data from pulsometer device and smart watch may be used; it dives into problems, related to selection of right regression model, in order to predict future possible user's pulse by certain timeframe; and suggest its implementation. Also, the following article covers the basic idea of classification in terms of detection specified condition associated with cardiovascular system based on aggregated data. And, finally, suggest its own approach of how to build analytics in long-running tasks. Keywords: linear regression, pulsometer, smart watch, heart rate, Javascript, fuzzy logic, machine learning, classification.
ДОЛГОСРОЧНЫЕ ПРЕДСКАЗАНИЯ В ЗАДАЧАХ, СВЯЗАННЫХ С ИЗМЕРЕНИЕМ ПУЛЬСА Абзалова Л.Р. (Российская Федерация)
Абзалова Лилия Радиковна - бакалавр, кафедра компьютерных систем и сетей, Высшая школа экономики, г. Москва
Аннотация: в данной статье рассматривается возможность использования собранных данных с пульсометров и умных часов; будут рассмотрены проблемы выбора регрессионной модели предсказаний возможного будущего пульса пользователя в разрезе времени и будет предложена собственная реализация решения для задач долгосрочного предсказания. Также в данной работе будет рассмотрена базовая идея применения методов классификации для выявления определенных состояний, связанных с сердечно-сосудистой системой, на основе собранных данных.
Ключевые слова: линейная регрессия, пульсометр, умные часы, сердечный ритм, Javascript, ленивые вычисления, машинное обучение, классификация.
1. INTRODUCTION
From ancient times up to our days, the humanity tries to simplify their lives. In struggle for technologies, the most common problem was transmitting and sharing of information. But, when the aim has been achieved, a new one challenge appeared - the data aggregation and interpretation.
One of these challenges is devoted to medicine problems, which is related to recognition of certain datasets (artefacts) and predictions, based on fetched data [1]. Among one of top priorities tasks in this sector - are researches, connected with heart rate administration[6]. This task includes not only right data interpretation, but also difficult aggregation and deep analytics, without taking in count certain out world parameters - like life style, stress and so on.
The suggested approach covers most of these aspects, in a brand new way - dynamic restricting of applied function for certain user. This will allow to calculate the average heart rate in a long run, by using a special coefficient over regression model, which is unique for everyone. Such solution should bring us closer in finding a cure from most diseases, related with the cardiovascular system, or at least, foresee some of them.
Regression models. Choosing model by nearest distribution
There is a large variety of suggested models for regression analytics. If we step back, and start from scratch, first question we ask - is the nature of our data, which we aggregate. In one or another case, by the nature of data, we can see the difference between total results of this or that applied function. For this reason, it is better to start from picking up the general functions among the suggested regressions functions, like linear regression, or SVM regression [4].
Building up decision model
In order to pick up the suitable model, we need to make sure, that the selected function's results are satisfiable enough. This is the general practice in supervised learning: we have a set of data, we train model with 80% of this data, and then test other 20% on this model [7]. The more accurate result - the better. For our purpose, we've decided to use python data science libs (in our case sklearn):
Models = []
models.append(('LR', LinearRegression())) models.append(('ELN', ElasticNet()))
models.append(('DT', DecisionTreeRegressor())) models.append(('SVM', SVR())) And append our test data to them: for name, model in models:
kfold = model_selection.KFold(n_splits=5, random_state=seed) cv_results = model_selection.cross_val_score(model, X_train, Y_train, cv=kfold, scoring=scoring)
results.append(cv_results) names.append(name) Results
LR: 0.849096 (0.073113) ELN: 0.652377 (0.108876) DT: 0.631982 (0.059994) SVM: 0.762597 (0.024189)
As you can see, the better correlation could be achieved by using simple linear regression. The second place takes SVM. This produce a feeling, that data distribution has soft edges, that is why SVM and LR have the highest accuracy than ELN and DT algorithms. Conclusion
In this article, we've discussed how do we treat with our datasets, picked up the right algorithm, and dived into applied statistics. In future work, we plan to talk about unification of function for certain user, find distribution error, and will try to a coefficient, which we will apply to regression model, in order to decrease the distribution error.
References in English / Список литературы на английском языке
1. Shouman Mai, Turner Tim, Stocker Rob. "Using Data Mining Techniques In Heart Disease Diagnoses And Treatment". Electronics, Communications and Computers (JECECC), 2012 Japan-Egypt Conference March, 2012. P. 173-177.
2. Detrano Robert, Janosi Andras, Steinbrunn Walter, Pfisterer Matthias, Schmid JohannJakob, Sandhu Sarbjit, Guppy Kern H., Lee Stella, Froelicher Victor, 1989 "International application of a new probability algorithm for the diagnosis of coronary artery disease" The American Journal of Cardiology. P. 304-310.
3. Gruhn N., Larsen F.S., Boesgaard S., Knudsen G.M., Mortensen S.A., Thomsen G., Aldershvile J. Cerebral blood flow in patients with chronic heart failure before and after heart transplantation. Stroke. 32, 2001. P. 2530-2533.
4. Polat K., Sahan S. and Gunes S. "Automatic detection of heart disease using an artificial immune recognition system (AIRS) with fuzzy resource allocation mechanism and k-nn (nearest neighbour) based weighting preprocessing". Expert Systems with Applications, 2007. P. 625-631.
5. Gopinathannair R., Mazur А. and Olshansky В. Syncope in congestive heart failure, J Cardiol. 15, 2008. P. 303-312.
6. Ozsen S., Gunes S., 2009 Attribute weighting via genetic algorithms for attribute weighted artificial immune system (AWAIS) and its application to heart disease and liver disorders problems. Expert Systems with Applications. P. 386-392.
7. Das Resul, Turkoglub Ibrahim and Sengurb Abdulkadir, 2009. "Effective diagnosis of heart disease through neural networks ensembles". Expert Systems with Applications. P. 767 -7680 [11]. Lamia Abed Noor Muhammed 2012 "Using Data Mining technique to diagnosis heart disease".
8. Lee C.W., Lee J.H., Lim T.H., Yang H.S., Hong M.K., Song J.K., Park S.W., Park S.J., Kim J. J., Prognostic significance of cerebral met.
9. Grodins F.S., Buell J. and Bart A.J. Mathematical analysis and digital simulation of the respiratory control system, J Appl Physiol. 22, 1967. P. 260-276.
10. Heldt Т. Computational Models of Cardiovascular Response to Orthostatic Stress. Ph.D. Thesis, Harvard/MIT Division of Health Science and Technology. MIT, 2004.
11. Khoo M.C.K., Gottschalk А. and Pack A.I. Sleep-induced periodic breathing and apnea: A theoretical study. J Appl Physiol. 70, 1991. P. 2014-2024.
Список литературы /References
1. Шуман Май, Тернер Тим, Стокер Роб. «Использование интеллектуального анализа данных в диагностике и лечении заболеваний сердца». Электроника, информация и компьютеры (JECECC), 2012. Японско-Египетская конференция. Март, 2012. С. 173-177.
2. Детрано Роберт, Яноси Андрес, Штейнбрун Уолтер, Фистер Матиас, Смит Джон-Якоб, Санду С., Гуппи Керн Х., Ли Стелла, Фроличер Виктор, 1989. «Международная программа возможностей нового алгоритма для диагностики заболеваний коронарной артерии». Американский журнал кардиологии. С. 304-315.
3. Грун Н., Ларсен Ф.С., Босгард С., Нудсен Г.М., Мортенсен С.А., Томсен Д., Альдершвиль Дж., «Мозговое кровообращение пациентов с хроническим пороком сердца, до и после трансплантации сердца». Инсульт. 32, 2001. С. 2530-2533.
4. Полат К., Сахан С. И Гинес С., 2007 «Автоматизированное определение заболеваний сердца с использованием искусственного распознавания иммунной системы (AIRS) с механизмом нечеткого распределения ресурсов и k-ная (ближайшие точки) основанной на определении нагрузки предварительной обработки» Экспертные системы с применением, 2007. С. 625-631.
5. Гопинатанер Р., Мазур А. и Ольшанский Б. Обморок в застойной сердечной недостаточности. Дж. Кардиол. 15, 2008. С. 303-312.
6. Озсен С., Гинес С., 2009 «Весовой множитель через генетический алгоритм для весового коэффициента для искуственной иммунной системы (AWAIS) и его применение для заболеваний сердца и заболеваний печени». Экспертные системы с применением. С. 386-392.
7. Дас Расул, Туркоглуб Ибрагим и Сенгурб Абдулхадир, 2009. «Эффективная диагностика заболеваний сердца через группу нейронных сетей». Экспертные системы с применением. С. 7675-7680.
8. Ли С.В., Ли Т.Х., Лим Т.Х., Янг Х.С., Хонг М.К., Сонг Дж.К., Парк С.В., Парк С.Дж., Ким Д.Д. Прогнозирование черепно-мозговой активности.
9. Гродинс Ф.С., Буэл Д. и Барт А.Дж. Математический анализ и цифровое моделирование органов дыхания. J Appl Physiol. 22, 1967. С. 260-276.
10. Хелт Т. Вычислительная модель ответа сердечно-сосудистой системы на ортостатическое давление,.Ph.D. Thesis, Harvard/MIT. Делегация естественных и прикладных наук для здоровья. MIT, 2004.
11. КхоМ.С.К., ГотсчалкА. и ПакА.И. Периодическое дыхание Чейна-Стокса и апноэ в искусственном сне: теоретическое исследование. J Appl Physiol. 70, 1991. С. 2014-2024.
