| fullrecord |
[{"key": "dc.contributor.advisor", "value": "Avela, Janne", "language": "", "element": "contributor", "qualifier": "advisor", "schema": "dc"}, {"key": "dc.contributor.advisor", "value": "Cronin, Neil", "language": "", "element": "contributor", "qualifier": "advisor", "schema": "dc"}, {"key": "dc.contributor.advisor", "value": "Rantalainen, Timo", "language": "", "element": "contributor", "qualifier": "advisor", "schema": "dc"}, {"key": "dc.contributor.advisor", "value": "K\u00f6ykk\u00e4, Miika", "language": "", "element": "contributor", "qualifier": "advisor", "schema": "dc"}, {"key": "dc.contributor.author", "value": "Vohlakari, Krista", "language": "", "element": "contributor", "qualifier": "author", "schema": "dc"}, {"key": "dc.date.accessioned", "value": "2022-06-06T10:18:50Z", "language": null, "element": "date", "qualifier": "accessioned", "schema": "dc"}, {"key": "dc.date.available", "value": "2022-06-06T10:18:50Z", "language": null, "element": "date", "qualifier": "available", "schema": "dc"}, {"key": "dc.date.issued", "value": "2022", "language": "", "element": "date", "qualifier": "issued", "schema": "dc"}, {"key": "dc.identifier.uri", "value": "https://jyx.jyu.fi/handle/123456789/81503", "language": null, "element": "identifier", "qualifier": "uri", "schema": "dc"}, {"key": "dc.description.abstract", "value": "T\u00e4m\u00e4n opinn\u00e4ytety\u00f6n tarkoituksena oli hy\u00f6dynt\u00e4\u00e4 koneoppimista kinemaattisten muuttujien arviointiin inertiaalisensoreista saatavasta datasta keih\u00e4\u00e4nheitossa. Perinteisten liikeanalyysimenetelmien rinnalle olisi hyv\u00e4 l\u00f6yt\u00e4\u00e4 menetelmi\u00e4, jotka mahdollistaisivat nopean palautteen antamisen urheilijoille. Datankeruun helpottuessa my\u00f6s aiempaa tarkempi ja suuremmalle kohderyhm\u00e4lle teht\u00e4v\u00e4 tutkimus mahdollistuisi.\n\nKymmenen tavoitteellisesti korkealla tasolla kilpailevaa keih\u00e4\u00e4nheitt\u00e4j\u00e4\u00e4 osallistui tutkimukseen. Mittaukset j\u00e4rjestettiin Kuortaneen urheiluopistolla sis\u00e4tiloissa kev\u00e4\u00e4ll\u00e4 2021. Urheilijat suorittivat tutkimusprotokollan osana tavallista harjoitteluaan. Suoritukset mitattiin optoelektronisella Vicon-j\u00e4rjestelm\u00e4ll\u00e4. T\u00e4m\u00e4 menetelm\u00e4 toimi \u201dkultaisena standardina\u201d muiden menetelmien vertailua varten. Vicon-j\u00e4rjestelm\u00e4\u00e4n yhdistettiin Bluetoothin avulla langattomasti BlueTrident-inertiaalisensorit, jotka kiinnitettiin heitt\u00e4j\u00e4n lantiolle, reisiin ja s\u00e4\u00e4riin.\n\nViidest\u00e4 sensorista saatuja lineaarikiihtyvyyksi\u00e4 ja kulmanopeuksia k\u00e4ytettiin pitk\u00e4kestoista ty\u00f6muistia hy\u00f6dynt\u00e4vien LSTM-neuroverkkomallien sy\u00f6ttein\u00e4. Vastemuuttujia olivat lantion orientaatio (tilt, list, rotation) sek\u00e4 lonkkien ja polvien kolmiulotteiset nivelkulmat. Ohjattua oppimista varten vastemuuttujat laskettiin Vicon-j\u00e4rjestelm\u00e4st\u00e4 saaduista markkerien koordinaateista avoimen l\u00e4hdekoodin OpenSim-ohjelmistolla. Kolme erilaista mallia koulutettiin. Yksi hy\u00f6dynsi jokaista viitt\u00e4 sensoria ja opetus tehtiin j\u00e4tt\u00e4en vuorotellen yksi urheilijoista testidataksi. Toiseen malliin k\u00e4ytettiin ainoastaan tukijalan reisi- ja s\u00e4\u00e4risensoria ja vastemuuttujana oli pelkk\u00e4 polvikulma. Kolmas malli k\u00e4ytti kaikkia viitt\u00e4 sensoria, mutta se oli urheilijakohtainen ja opetus tehtiin j\u00e4tt\u00e4en vuorotellen yksi maksimaalinen heitto testidataksi.\n\nUrheilijakohtainen malli ennusti kaikki muuttujat parhaiten (RMSE, 2\u20135.4 astetta; ICC1, 0.88\u20130.98), mutta oli luultavasti ylisovittunut datan v\u00e4hyyden vuoksi. Viiden sensorin ryhm\u00e4mallit ennustivat muuttujat RMSE:n ollen keskim\u00e4\u00e4rin 7.7\u201316.6 astetta eri muuttujissa. Ennustetut taaemman jalan polvikulmat ja lantion kallistuminen sivuttaissuunnassa vastasivat huonosti \u201dkultaisen standardin\u201d antamia lukemia sis\u00e4korrelaatiokertoimella arvioituna (ICC1). Muut muuttujat pystyttiin ennustamaan keskim\u00e4\u00e4r\u00e4isesti tai hyvin (ICC1, 0.72\u20130.87). Kahden sensorin polvikulmamalli oli parempi kuin viiden sensorin ryhm\u00e4malli ja huonompi kuin urheilijakohtainen malli. Ryhm\u00e4mallit toimivat yll\u00e4tt\u00e4v\u00e4n hyvin datan v\u00e4hyyteen verraten. Mik\u00e4li dataa ohjattua oppimista varten olisi tarpeeksi, voisi olla kannattavaa testata tiettyihin keih\u00e4\u00e4nheittosuorituksen vaiheisiin perustuvan LSTM-mallin yleistyvyyskyky\u00e4.", "language": "fi", "element": "description", "qualifier": "abstract", "schema": "dc"}, {"key": "dc.description.abstract", "value": "The aim of this thesis was to use machine learning to predict kinematic variables in javelin throwing using IMU data. In addition to traditional motion analysis methods, new more flexible methods would enable faster feedback to athletes. If data collection could be simplified, it would be possible to collect accurate data from larger populations.\n\nTen well trained international and national level athletes participated in the study. Measurements were conducted at Kuortane Olympic Training Center in spring 2021. Athletes performed the study protocol as part of their general training. Javelin throw performances were captured via an optoelectronic Vicon system, which was used as the \u201cgold standard\u201d for comparisons to other methods. BlueTrident inertial measurement units were connected wirelessly via Bluetooth to the Vicon system. Sensors were mounted to the pelvic area near the sacrum, thighs and shanks.\n\nLinear accelerations and angular velocities from five sensors were used as inputs for long short-term memory (LSTM) models. Pelvic orientations (tilt, list, rotation) and bilateral hip and knee angles were used as target variables. For supervised learning, the target variables were calculated with open source OpenSim software using marker trajectories recorded via Vicon. Three different LSTM models were trained. The first used data from all five sensors and training was implemented using the leave one out method. The second model used only two sensors mounted to the brace leg, with the goal of predicting only the knee angle of this leg. The third model used all five sensors but was athlete-specific, and training was again implemented using the leave one out method.\n\nAthlete-specific individual models yielded the most accurate predictions for all variables (RMSE 2\u20135.4 degrees, ICC1 0.88\u20130.98) but these models were probably overfitted due to the small training dataset. For the five sensor group model, the RMSEs were 7.7\u201316.6 degrees for all variables. Based on ICC1, the knee angles of the rear leg and pelvic list corresponded poorly with the \u201cgold standard\u201d. Other variables were predicted with moderate or good accuracy (ICC1 0.72\u20130.87). The two sensor knee angle model performed better than the five sensor group model but worse than athlete-specific individual models. Group models worked surprisingly well considering the small dataset. If a larger dataset were available for supervised learning, it would be valuable to explore the generalisability of the LSTM model for predicting biomechanical parameters at certain phases of the javelin throw.", "language": "en", "element": "description", "qualifier": "abstract", "schema": "dc"}, {"key": "dc.description.provenance", "value": "Submitted by Miia Hakanen (mihakane@jyu.fi) on 2022-06-06T10:18:50Z\nNo. of bitstreams: 0", "language": "en", "element": "description", "qualifier": "provenance", "schema": "dc"}, {"key": "dc.description.provenance", "value": "Made available in DSpace on 2022-06-06T10:18:50Z (GMT). No. of bitstreams: 0\n Previous issue date: 2022", "language": "en", "element": "description", "qualifier": "provenance", "schema": "dc"}, {"key": "dc.format.extent", "value": "67", "language": "", "element": "format", "qualifier": "extent", "schema": "dc"}, {"key": "dc.format.mimetype", "value": "application/pdf", "language": null, "element": "format", "qualifier": "mimetype", "schema": "dc"}, {"key": "dc.language.iso", "value": "eng", "language": null, "element": "language", "qualifier": "iso", "schema": "dc"}, {"key": "dc.rights", "value": "In Copyright", "language": "en", "element": "rights", "qualifier": null, "schema": "dc"}, {"key": "dc.subject.other", "value": "inertial measurement unit", "language": "", "element": "subject", "qualifier": "other", "schema": "dc"}, {"key": "dc.title", "value": "Using Neural Networks to Estimate Lower Body Kinematics from IMU Data in Javelin Throwing", "language": "", "element": "title", "qualifier": null, "schema": "dc"}, {"key": "dc.type", "value": "master thesis", "language": null, "element": "type", "qualifier": null, "schema": "dc"}, {"key": "dc.identifier.urn", "value": "URN:NBN:fi:jyu-202206063120", "language": "", "element": "identifier", "qualifier": "urn", "schema": "dc"}, {"key": "dc.type.ontasot", "value": "Pro gradu -tutkielma", "language": "fi", "element": "type", "qualifier": "ontasot", "schema": "dc"}, {"key": "dc.type.ontasot", "value": "Master\u2019s thesis", "language": "en", "element": "type", "qualifier": "ontasot", "schema": "dc"}, {"key": "dc.contributor.faculty", "value": "Liikuntatieteellinen tiedekunta", "language": "fi", "element": "contributor", "qualifier": "faculty", "schema": "dc"}, {"key": "dc.contributor.faculty", "value": "Faculty of Sport and Health Sciences", "language": "en", "element": "contributor", "qualifier": "faculty", "schema": "dc"}, {"key": "dc.contributor.department", "value": "Liikunta- ja terveystieteet", "language": "fi", "element": "contributor", "qualifier": "department", "schema": "dc"}, {"key": "dc.contributor.department", "value": "Sport and Health Sciences", "language": "en", "element": "contributor", "qualifier": "department", "schema": "dc"}, {"key": "dc.contributor.organization", "value": "Jyv\u00e4skyl\u00e4n yliopisto", "language": "fi", "element": "contributor", "qualifier": "organization", "schema": "dc"}, {"key": "dc.contributor.organization", "value": "University of Jyv\u00e4skyl\u00e4", "language": "en", "element": "contributor", "qualifier": "organization", "schema": "dc"}, {"key": "dc.subject.discipline", "value": "Biomekaniikka", "language": "fi", "element": "subject", "qualifier": "discipline", "schema": "dc"}, {"key": "dc.subject.discipline", "value": "Biomechanics", "language": "en", "element": "subject", "qualifier": "discipline", "schema": "dc"}, {"key": "yvv.contractresearch.collaborator", "value": "public", "language": "", "element": "contractresearch", "qualifier": "collaborator", "schema": "yvv"}, {"key": "yvv.contractresearch.funding", "value": "0", "language": "", "element": "contractresearch", "qualifier": "funding", "schema": "yvv"}, {"key": "yvv.contractresearch.initiative", "value": "student", "language": "", "element": "contractresearch", "qualifier": "initiative", "schema": "yvv"}, {"key": "dc.type.coar", "value": "http://purl.org/coar/resource_type/c_bdcc", "language": null, "element": "type", "qualifier": "coar", "schema": "dc"}, {"key": "dc.rights.accesslevel", "value": "openAccess", "language": null, "element": "rights", "qualifier": "accesslevel", "schema": "dc"}, {"key": "dc.type.publication", "value": "masterThesis", "language": null, "element": "type", "qualifier": "publication", "schema": "dc"}, {"key": "dc.subject.oppiainekoodi", "value": "5012", "language": "", "element": "subject", "qualifier": "oppiainekoodi", "schema": "dc"}, {"key": "dc.subject.yso", "value": "teko\u00e4ly", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "keih\u00e4\u00e4nheitto", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "koneoppiminen", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "liikeoppi", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "artificial intelligence", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "javelin throwing", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "machine learning", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.subject.yso", "value": "kinematics", "language": null, "element": "subject", "qualifier": "yso", "schema": "dc"}, {"key": "dc.format.content", "value": "fulltext", "language": null, "element": "format", "qualifier": "content", "schema": "dc"}, {"key": "dc.rights.url", "value": "https://rightsstatements.org/page/InC/1.0/", "language": null, "element": "rights", "qualifier": "url", "schema": "dc"}, {"key": "dc.type.okm", "value": "G2", "language": null, "element": "type", "qualifier": "okm", "schema": "dc"}]
|