Bahuguna Abhinav, Uniyal Akanksha, Vallabh Vidisha
Department of Biostatistics, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, Uttarakhand, India.
Department of Community Medicine, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, India.
BMC Public Health. 2025 Aug 7;25(1):2693. doi: 10.1186/s12889-024-21073-9.
Infant mortality is an important predictor of a government's commitment to its people. Global infant deaths have declined since past decades but at a pace that leaves much to be desired. India's declining pattern of trends is encouraging but the low performance of individual states like Madhya Pradesh (MP) indicates an urgent need for policy revision and implementation.
This paper forecasts the Infant mortality rate (IMR) of India and MP by the year 2030 through autoregressive integrated moving average (ARIMA) model after obtaining stationarity by differencing the series of IMR once. The Akaike's information criterion and Bayesian information criterion have been used for the selection of best ARIMA model amongst other existing choices. The model diagnostics through Ljung and Box test shows absence of autocorrelation in the residuals (p > 0.05).
The findings through ARIMA(3, 1, 0) foretell a declining IMR from 27 to 20 per thousand live births (from 2021 to 2030) in India. Similarly, MP is expected to experience reduction in infant deaths from 44 to 39 per thousand live births (from 2021 to 2030). The deployed model is well fitted as mean absolute percentage error lies below 5%. During 2010-20, India and MP witnessed a decadal reduction of 40% and 31% in IMR, respectively. From the year 2010 onwards, India experienced the highest annual reduction of 8.1% in IMR during 2015-16. Similarly, MP encountered a decrease of 6.5% in IMR recently (2019-20), which is the highest declining annual IMR in the state during past ten years.
The projected figures on IMR are satisfactory for policy makers at national level, but MP is still miles away to achieve acceptable IMR as compared to the country's IMR. The state requires more attention and focus on exploring reasons and identifying underlying factors responsible for higher IMR across its demographic structure including socio-economic characteristics.
婴儿死亡率是衡量政府对民众承诺的一项重要指标。在过去几十年里，全球婴儿死亡人数有所下降，但下降速度仍不尽人意。印度婴儿死亡率呈下降趋势，这令人鼓舞，但像中央邦这样个别邦的低表现表明迫切需要修订和实施相关政策。
本文通过对婴儿死亡率序列进行一次差分以使其平稳后，利用自回归积分移动平均（ARIMA）模型预测印度和中央邦到2030年的婴儿死亡率。赤池信息准则和贝叶斯信息准则被用于在其他现有选择中挑选最佳的ARIMA模型。通过Ljung和Box检验进行的模型诊断表明残差不存在自相关性（p>0.05）。
通过ARIMA(3, 1, 0)模型得出的结果预测，印度每千例活产婴儿的死亡率将从27降至20（从2021年到2030年）。同样，中央邦预计每千例活产婴儿的死亡人数将从44降至39（从2021年到2030年）。所采用模型的拟合效果良好，平均绝对百分比误差低于5%。在2010 - 2020年期间，印度和中央邦的婴儿死亡率分别出现了40%和31%的十年降幅。从2010年起，印度在2015 - 2016年期间婴儿死亡率年度降幅最高，达到8.1%。同样，中央邦最近（2019 - 2020年）婴儿死亡率下降了6.5%，这是该邦过去十年中年度降幅最高的一次。
预测的婴儿死亡率数据对国家层面的政策制定者来说是令人满意的，但与全国婴儿死亡率相比，中央邦要实现可接受的婴儿死亡率仍有很大差距。该邦需要更多关注并着重探究原因，找出其人口结构（包括社会经济特征）中导致婴儿死亡率较高的潜在因素。
