目前治療股骨近端骨折最廣泛應用的固定器是標準迦瑪骨釘(Standard Gamma Nail, SGN)和動態髖骨螺絲(Dynamic Hip Screw, DHS)，但他們卻不是想像中的完美。從醫學臨床觀察中發現，在治療過程中植入物會有破壞和鬆脫的危險情況發生，而且迦瑪骨釘的末端會造成近端股骨有應力集中的現象，使會骨頭發生骨切效應(Cut-out)，近為了改善迦瑪骨釘以及動態髖骨螺絲的缺點，因而設計製造出雙螺絲骨釘(Double Screw Nail, DSN)。雙螺絲骨釘的結構包含一支骨髓內釘、兩支遲滯螺絲和一支或兩支的遠端鎖緊螺絲，雖然雙螺絲骨釘改善了珈瑪骨釘和動態髖骨螺絲的缺點，且擁有較佳的臨床性能，但其不臻完美，遲滯螺絲在術後短期內有破壞的情況發生，因此有必要深入研究遲滯螺絲的最佳化。
本研究將利用有限元素法用最壞情況(worst-case)來模擬螺絲的破壞情況，並套用田口品質工程法來探討遲滯螺絲中最重要的參數，後期將合併類神經網路與遺傳演算法來獲得最佳設計。
依照田口直交表(Orthogonal Array)規律的分配每一個模型的參數水準，然後，利用SolidWorks 2008建立立體模型，將其利用parasolid檔案轉進ANSYS Workbench 11中進行有限元素分析模擬。彎曲強度(bending strength)模型是記錄其最大張應力來作為目標值；而抗拉出強度(pullout strength)則是記錄總反作用力。之後，利用田口品質工程法裡的公式得到變異數分析(ANOVA)，從中比較每個設計因子的貢獻度。其後，將兩組數據(從彎曲模擬與咬合模擬中得到)輸入進類神經網路程式中運算出網路模型，並利用遺傳演算法找出最佳化的範圍。
本研究結果顯示，單獨針對彎曲強度而言，圓錐起始位置與內徑為其重要的設計參數，若針對抗拉出強度而言，內徑、近端根部弧角半徑及節距是重要的設計參數；在兩種強度性質之類神經網路模型的學習平均誤差與測試平均誤差均相當小(3%以內)，最大誤差也不超過5%，兩種網路模型結果與有限元素結果的相關係數都相當接近1；經遺傳演算法的最佳化演算後，圓錐起始位置的最佳值為0㎜、內徑為3.684㎜∼3.958㎜、近端根部弧角半徑為0.4㎜∼0.458㎜、節距為2.6㎜、近端傾斜半角為5°、螺牙厚度為0.126㎜∼0.132㎜。

The Trochanteric Gamma nail, TGN and the Dynamic Hip Screw, DHS have been widely used in the treatment of proximal femoral fractures. In clinical reports, the implants may fail or loosen during the treatment. It may cause severe problems for patients. In addition, the tip of TGN may cause the postoperative fractures because of the stress concentration. To decrease defects of two implants, Double Screw Nail, DSN was designed and manufactured. The structure of DSN included a consisted of a nail, two lag-screws, and one or two distal locking screws. Although DSN can decrease the defects from SGN and DHS and get a better clinical performance. But it was not perfect, either. We could observe the lag-screw failed in early treatment. Therefore, it is necessary to optimize the lag-screw.
The worst case was performed by use of Finite Element Method in the simulation of this study and combined Taguchi method to probe which parameter was the most significant. Finally, the combination of Artificial Neuron Network, ANN and Genetic algorithm, GA was used to get the optimization.
Following the regulations of Orthogonal Array table to assign every factors and those levels. And then created the 3-D models by using commercial software SolidWorks 2008 and imported models saved as parasolid files into commercial software ANSYS Workbench. The maximum tensile stress was recorded as the targets of bending strength models and total reaction force for pullout strength models. Afterward, Taguchi methods was applied to figure out analysis of the variance, ANOVA. In ANOVA table, we can get contributions of each factors. Employing ANN to develop the net models and using GA to find out the optimal scope.
The results of this study shows:
a.) The Initial Position of Conical angle is the most significant and the second one is Inner Diameter for bending strength model.
b.) For pullout strength model, the Inner Diameter, Proximal Root Radius and Pitch are important design factors.
c.) In ANN section, the average errors of learning and testing are quite low, the maximum error less than 5%, and the correlation coefficients between FEM results and ANN results are approach to 1 for both two strengths.
d.) In GA section, the optimal values or ranges are 0㎜for IP, 3.684㎜∼3.958㎜ for ID, 0.4㎜∼0.458㎜ for PRR, 2.6㎜ for Pi, 5° for PHA and 0.126㎜∼0.132㎜ for TW.

[1] 孫施盛, "退化性關節炎與骨質疏鬆性股骨頭之生物力學特性分析," 陽明大學醫學工程系碩士論文, (2002).
[2] 楊榮森, "臨床股骨學," 台北, 合記圖書出版社, p. 268, (1989).
[3] Kiefer, H., "Simultaneous transarticular intramedullary nailing of the tibia and total knee replacement in tibial nonunion and osteoarthrosis," Unfallchirurg, vol. 102, pp. 888-892, 1999.
[4] Gaine, W.J., Sanville, P.R., and Bamford, D.J., "The Charnley-Hastings bipolar prosthesis in femoral neck fractures - a study of dynamic motion," Injury-International Journal of the Care of the Injured, vol. 31, pp. 257-263, (2000).
[5] Halder, S.C., "The Gamma Nail for Peritrochanteric Fractures," J Bone Joint Surg, vol. 74B, pp. 340-344, (1992).
[6] 陳育彬, "股骨鎖定內釘之有限元素分析與生物力學測試," 台灣科技大學機械工程系碩士論文, (2001).
[7] Haynes, R.C., Poll, R.G., Miles, A.W., and Weston, R.B., "An experimental study of the failure modes of the Gamma Locking Nail and AO Dynamic Hip Screw under static loading: a cadaveric study," Medical Engineering & Physics, vol. 19, pp. 446-453, (1997).
[8] Bridle, S.H., Patel, A.D., Bircher, B., and Calvert, P.T., "fixation of intertrochanteric fractures of the femur," J Bone Joint Surg, vol. 73B, pp. 330-334, (1991).
[9] 周建基, "股骨近心端治療之有限元素分析," 台灣科技大學機械工程系碩士論文, (2004).
[10] "http://www.medscape.com."
[11] "http://www.aafp.org."
[12] Jenkins, G.W., Kemnitz, C.P., and Tortora, G.J., "Anatomy and physiology : from science to life " New Yourk, p. 235, (2007).
[13] "http://www.xxmu.edu.cn."
[14] Gaebler, C., Stanzl-Tschegg, S., Heinze, G., Holper, B., Milne, T., Berger, G., and Vecsei, V., "Fatigue strength of locking screws and prototypes used in small-diameter tibial nails: A biomechanical study," Journal of Trauma-Injury Infection and Critical Care, vol. 47, pp. 379-384, (1999).
[15] Youssef, J.A., McKinley, T.O., Yerby, S.A., and McLain, R.F., "Characteristics of pedicle screw loading - Effect of sagittal insertion angle on intrapedicular bending moments," Spine, vol. 24, pp. 1077-1081, Jun (1999).
[16] Lin, J., Lin, S.J., Chiang, H.S., and Hou, S.M., "Bending strength and holding power of tibial locking screws," Clinical Orthopaedics and Related Research, pp. 199-206, Apr (2001).
[17] Chen, P.Q., Lin, S.J., Wu, S.S., and So, H., "Mechanical performance of the new posterior spinal implant: Effect of materials, connecting plate, and pedicle screw design," Spine, vol. 28, pp. 881-886, (2003).
[18] Chao, C.K., Hsu, C.C., Wang, J.L., and Lin, J., "Increasing bending strength of tibial locking screws: Mechanical tests and finite element analyses," Clinical Biomechanics, vol. 22, pp. 59-66, Jan (2007).
[19] Chao, C.K., Hsu, C.C., Wang, J.L., and Lin, J., "Increasing bending strength and pullout strength in conical pedicle screws: Biomechanical tests and finite element analyses," Journal of Spinal Disorders & Techniques, vol. 21, pp. 130-138, Apr (2008).
[20] Abshire, B.B., McLain, R.F., Valdevit, A., and Kambic, H.E., "Characteristics of pullout failure in conical and cylindrical pedicle screws after full insertion and back-out," The Spine Journal, vol. 1, pp. 408-414, (2001).
[21] Ono, A., Brown, M.D., Latta, L.L., Milne, E.L., and Holmes, D.C., "Triangulated pedicle screw construct technique and pull-out strength of conical and cylindrical screws," (2001), pp. 323-329.
[22] Hou, S.M., Hsu, C.C., Wang, J.L., Chao, C.K., and Lin, J., "Mechanical tests and finite element models for bone holding power of tibial locking screws," Clinical Biomechanics, vol. 19, pp. 738-745, Aug (2004).
[23] Hsu, C.C., Chao, C.K., Wang, J.L., Hou, S.M., Tsai, Y.T., and Lin, J., "Increase of pullout strength of spinal pedicle screws with conical core: biomechanical tests and finite element analyses," Journal of Orthopaedic Research, vol. 23, pp. 788-794, Jul (2005).
[24] Davim, J.P., "An experimental study of the tribological behaviour of the brass/steel pair," Journal of Materials Processing Technology, vol. 100, pp. 273-277, Apr (2000).
[25] Davim, J.P., "Design of optimisation of cutting parameters for turning metal matrix composites based on the orthogonal arrays," Journal of Materials Processing Technology, vol. 132, pp. 340-344, Jan (2003).
[26] Agatonovic-Kustrin, S. and Beresford, R., "Basic concepts of artificial neural network (ANN) modeling and its application in pharmaceutical research," Journal of Pharmaceutical and Biomedical Analysis, vol. 22, pp. 717-727, Jun (2000).
[27] Dar, F.H., Meakin, J.R., and Aspden, R.M., "Statistical methods in finite element analysis," Journal of Biomechanics, vol. 35, pp. 1155-1161, Sep (2002).
[28] 葉怡成, "類神經網路模式應用與實作," 儒林出版社, (2001).
[29] Vellido, A., Lisboa, P.J.G., and Vaughan, J., "Neural networks in business: a survey of applications (1992-1998)," Expert Systems with Applications, vol. 17, pp. 51-70, Jul 1999.
[30] Hornik, K., Stinchcombe, M., and White, H., "Multilayer feedforward networks are universal approximators," Neural Networks, vol. 2, pp. 359-366, (1989).
[31] 林萍珍, "投資分析(含Matlab應用、類神經網路與遺傳演算法模型)," 新陸書局, pp. 520-526, (2008).
[32] Hsu, C.C., Chao, C.K., Wang, J.L., and Lin, J., "Multiobjective optimization of tibial locking screw design using a genetic algorithm: Evaluation of mechanical performance," Journal of Orthopaedic Research, vol. 24, pp. 908-916, May (2006).
[33] Kwok, A.W.L., Finkelstein, J.A., Woodside, T., Hearn, T.C., and Hu, R.W., "Insertional torque and pull-out strengths of conical and cylindrical pedicle screws in cadaveric bone," Spine, vol. 25, pp. 19S-24S, Mar (2000).
[34] 徐慶琪, "骨螺絲之結構設計與生物力學分析," 台灣科技大學機械工程系博士論文, (2005).
[35] Basheer, I.A. and Hajmeer, M., "Artificial neural networks: fundamentals, computing, design, and application," Journal of Microbiological Methods, vol. 43, pp. 3-31, Dec (2000).