Diagnosis Support by Machine Learning Using Posturography DataTeruKamogashira
?
Machine learning algorithms can help analyze posturography data to diagnose vestibular dysfunction. An evaluation of various algorithms found that gradient boosting had the best performance with an AUC of 0.90. While deep learning did not perform best, optimizing algorithm parameters is important. Larger, multi-institutional clinical datasets may improve machine learning's ability to accurately diagnose vestibular disorders from posturography data.
Responses from the trapezoid body in the Mongolian gerbilTeruKamogashira
?
The study recorded responses from 80 fibers in the trapezoid body of the Mongolian gerbil. 26 fibers responded best to sounds in the ipsilateral ear and 54 to the contralateral ear. Many onset responses were observed, which is unusual compared to other mammals like cats. Onset responses occurred over a similar depth as primary-like responses. This suggests more diversity in response types of neurons in the gerbil anteroventral cochlear nucleus than in cats.
Diagnosis Support by Machine Learning Using Posturography DataTeruKamogashira
?
Machine learning algorithms can help analyze posturography data to diagnose vestibular dysfunction. An evaluation of various algorithms found that gradient boosting had the best performance with an AUC of 0.90. While deep learning did not perform best, optimizing algorithm parameters is important. Larger, multi-institutional clinical datasets may improve machine learning's ability to accurately diagnose vestibular disorders from posturography data.
Responses from the trapezoid body in the Mongolian gerbilTeruKamogashira
?
The study recorded responses from 80 fibers in the trapezoid body of the Mongolian gerbil. 26 fibers responded best to sounds in the ipsilateral ear and 54 to the contralateral ear. Many onset responses were observed, which is unusual compared to other mammals like cats. Onset responses occurred over a similar depth as primary-like responses. This suggests more diversity in response types of neurons in the gerbil anteroventral cochlear nucleus than in cats.
10. Peak amplitude と Peak
latency について、
方法( New vs.
Conventional)
刺激間隔 (Short vs. Long)
の効果を 2 way repeated
measures ANOVA で検定し
た
方法の主効果は見られなかった
が、刺激間隔に関しては
Short(1.0s) の方が Long(3.0s)
よりも有意に振幅が減衰してい
た ( 右半球で p=0.014 、左半球
で p=0.012 )
潜時についてはいずれの効果も
見られなかった。
11. 測定時間
New paradigm では100回加算の Short 1回と Long 1回であわせ
て20分
Conventional oddball paradigm では25回加算の Short ( Tone 1
回と Phoneme 1回)と Long ( Tone 1回と Phoneme 1回)で計
4回で約30分
加算回数4倍になり、測定時間は短縮された。