جستجو
Probabilistic modeling of disease progression
Elahe Dorani1 Zahra Narimani2 Holger Froehlich3
Abstract :
Disease progression modeling is the use of mathematical and statistical approaches for tracking the disease severity in a special period of timecite{cook2016disease}. One of the major problems in medicine is detecting the chronic diseases in the first stages of progression. Chronic disease like Alzheimer s start from mild symptoms and progress slowly to the severe stages during a long time.
Early detection of symptoms leads to more effective therapies, detect different subtypes of disease, and designing more effective drugs. par
In previous research, they have used limited well-known features for modeling progression of Alzheimer s disease and algorithms such as Random Forestscite{dimitriadis2018random, moore2019random}, SVMscite{kloppel2008automatic, zhang2011multimodal}, and Linear Mixed-Effects modelscite{sabuncu2011dynamics} are hired to distinguish the healthy elderly, patients with mild or sever impairments and AD.par
In this work we used an automatic feature selection method on high dimensional data set and controlling the complexity of the model as well. We focused on probabilistic modeling of Alzheimer’s disease progression using TADPOLE data set. This data set contains different groups of features, such as demographics, cognitive scores, genetic bio-markers, MRI, DTI and PET scans. We applied Bayesian approach using variational inference and sparse prior.par
We compared our model with the reference model proposed in cite{Zhu2018APD} using normal prior, the Random Forests proposed in cite{moore2019random} and Linear Mixed-Effects regression. The target variables that we were interested in are Ventricles, Hippocampus, MMSE, ADAS13 and CDRSB as the three well-known cognitive tests. In the first experiment, comparing the results of proposed Bayesian model using sparse prior (Laplace distribution) with the reference model using normal priorcite{Zhu2018APD}, showed us that that it was more accurate for the Ventricles, Hippocampus, ADAS13 and CDRSB than the reference model.
In the second experiment we used the Random Forest both as the features selection method for the proposed Bayesian model and benchmark. As another comparison, this version of the Bayesian model with the reference model showed us that on it was more accurate on Ventricles, Hippocampus, ADAS13 and CDRSB than the reference model. But the LME model and Random Forests regression algorithm were better than the reference and proposed models.
Keywords : disease progression modeling bayesian model variational inference sparse prior bayesian feature selection
زبان مقاله : انگلیسی
محور مقاله : بیو انفورماتیک
محل انتشار : دومین همایش ملی ریاضیات زیستی
آدرس وب سایت کنفرانس : biomathtabrizu.ir
کد IOI مقاله : XCBD-FZDHG
لینک مستقیم مقاله : https://isnac.ir/XCBD-FZDHG
نمایش چکیده فارسی
: https://isnac.ir/SA/XCBD-FZDHG
سایر مقالات ارائه دهنده
سایر مقالات محل انتشار
- Zero-Coupon Bond Pricing under the double Hull-White Model
- مدل سازی بیماری کووید-19
- Probabilistic modeling of disease progression
- On the conformable fractional differential equation of human liver model
- تعیین بهترین تابع ریاضی توصیف کنندهی منحنی رشد در بزهای نژاد عدنی
- کاربرد آمار زیستی در روانشناسی
- Multiple bifurcations analysis in a delayed predator-prey system with disease in prey and stage structure for predator
- استفاده ازAHP فازی برای بررسی تاثیر یوگا بر کیفیت زندگی مبتلایان به سرطان پستان
- Mathematical model of tuberculosis transmission in a two-strain with therapy
- استنباط بیزی شبکههای تنظیم ژنی
- The numerical solution of Volterra s population model Based on the collocation method and guassian functions
- برآورد مدل های پویا برای شبکه های تنظیم ژن
- Phase plane analysis of T-type calcium channel
- Polynomial of antioxidant activity related to peanut based on different salinity levels
- کاربرد ریاضیات در بوم شناسی