msm(formula = sarcopenia_status ~ follow_year, subject = ID,     data = hb6, qmatrix = matrixQ, covariates = ~BMI, method = "BFGS",     control = list(fnscale = 4000, maxit = 10000))Optimisation probably not converged to the maximum likelihood.optim() reported convergence but estimated Hessian not positive-definite.-2 * log-likelihood:  10706.39

想问一下这种情况怎么解决