Preliminary Construction Of A Fall Risk Prediction Model For Elderly Females Based On Discriminant Analysis

Objective: To explore the sensitive index and influence weight of fall occurrence,enrich the multi-factor research system of fall risk,and provide more experimental basis for fall prevention,risk prediction and healthy old-age care.Methods :128 healthy elderly women aged 60-70 years in Tianjin were selected.According to the investigation of fall history,they were divided into fall group and non-fall group,and the body function indexes such as body composition,lower limb muscle strength,dynamic and static balance ability and walking gait characteristics were measured respectively.Using independent sample t to test the difference of index between statistical groups,paired sample t to test the difference of index within statistical group,select the index of difference between groups into the discriminant analysis model,establish a typical discriminant function,and verify the validity of the model with internal verification,cross verification and external test.Results :1.Comparison of body function indicators:(1)Body composition contrast results showed that the skeletal muscle content of fall group was significantly lower than that of non-fall group(P<0.05).(2)Comparative results of sensory function showed that the visual acuity was significantly lower than that of the non-fall group(P<0.01)and the knee joint active position in the fall group(P<0.01).(3)The results showed that :(1)Korebalance test: the dynamic track FBR of standing with one-legged open eyes in fall group was significantly higher than that in non-fall group(P<0.05);(2)simple test: the standing time of one-legged closed eyes in fall group was significantly lower than that of non-fall group(P<0.01);the walking time of eight feet in fall group was significantly lower than that of non-fall group(P<0.05).(4)Comparative results of lower limb muscle strength showed:(1)Isokinetic testing: The APM of flexor of nondominant lateral malleolus in fall group was significantly lower than that in non-fall group(P<0.01),The APM of nondominant flexor of knee joint APM、flexor of bilateral knee joint in fall group was significantlylower than that in non-fall group(P<0.05);(2)simple test: the number of sitting up for30 seconds in the fall group was significantly smaller than that in the non-fall group(P<0.05).(5)Comparison of gait characteristics:(1)Gait cycle parameters: the mean stride speed of fall group was significantly smaller than that of non-fall group(P<0.05);(2)Kinematics parameters: Non-dominant lateral malleolus angle was significantly smaller than that of non-fall group(P<0.05),non-dominant lateral hip angle,non-dominant lateral malleolus angle and lateral malleolus angle were significantly smaller than those of non-fall group(P<0.01);At the moment of contact,the lateral malleolus angle was significantly larger than that of the nondominant side(P<0.05),Both the dominant knee angle and hip angle were significantly larger than the nondominant side(P<0.01),and the dominant side knee angle was significantly larger than that of the non-dominant side in the non-fall group(P<0.05).2.The typical discriminant function and Bayesian classification function are established respectively,The index weight size and order included in the discriminant model are: Moment left ankle Angle from the ground(Dr = 0.575)> moment right ankle Angle from the ground(Dr = 0.422)> knee active position sense(Dr = 0.416)>30 seconds sitting up number(Dr = 0.409)> 10% gray vision(Dr = 0.374)> under one leg stand close their eyes(Dr = 0.359)> eight feet stand walk time(Dr = 0.348)>touchdown moment left ankle Angle(Dr = 0.330)> content of skeletal muscle(Dr =0.331)> bilateral knee flexors APM ratio(Dr = 0.281)> average pace(Dr = 0.274)>left hip Angle(Dr =0.256)> non-dominant side ankle flexor peak moment(Dr=0.216)> non-dominant side knee flexor APM(Dr =0.213)> dynamic trajectory FBR(Dr =-0.197)> single leg open eyes standing RLR)(Dr =-0.053).3.The preliminary construction of the model was completed successfully,and the prediction of the discriminant function model was good.The correct prediction rates of self-verification,cross-verification and external verification were 98.4%,93.8% and 96.4%,respectively.Conclusions: 1.completed the pre-judgment model construction of elderly women,and the model has good predictability after testing.The function modelshows that the test indexes which can better reflect the fall risk of elderly women are left and right ankle angle,knee joint active position sense,30 seconds sitting up number,10% gray level vision,single leg closed eye standing time.2.should simultaneously improve the gait stability and walking function of elderly women walking.The balance function level of the elderly women in the closed state is more important,and the combination of static and dynamic training,open eyes and closed eyes should be adopted in the prevention of falls.Improving the balance of the strength of the lower limbs and the coordination of the joints of the elderly women is the self-protection factor to prevent falls.