Deszczyński J
Oddział Chirurgii Urazowo-Ortopedycznej, Szpital Czerniakowski w Warszawie.
Chir Narzadow Ruchu Ortop Pol. 1999;64(6):649-56.
This paper focuses on heuristic techniques based on artificial intelligence namely artificial non-linear and multilayer neurons employed for assessment of bone union in fractures treated by Dynastab DK external fixation. The software used allowed for prediction of bone union curve at the early stage of treatment. The training of the neural net has been done on the ground of 56 cases of Dynastab DK treated fractures. The net served for periodical assessment of bone union and its further prediction in 17 long bone shaft fractures. Bone mechanical properties at the fracture site showed non-linear changes in time. Specifically great changes were noted 4 months postoperatively. Flexion at the fracture site is critical. Neuron net proved to be useful in predicting bone union curve.
本文着重探讨基于人工智能的启发式技术,即采用人工非线性和多层神经元来评估使用Dynastab DK外固定治疗的骨折的骨愈合情况。所使用的软件能够在治疗早期预测骨愈合曲线。神经网络的训练基于56例接受Dynastab DK治疗的骨折病例。该网络用于对17例长骨干骨折的骨愈合进行定期评估及其进一步预测。骨折部位的骨力学性能随时间呈非线性变化。具体而言,术后4个月观察到显著变化。骨折部位的弯曲情况至关重要。神经网络在预测骨愈合曲线方面被证明是有用的。