ABSTRACT
The Zimbabwean pharmaceutical industry is made up of eight generic pharmaceutical manufacturing companies for finished human medicines. Statistically, the local pharmaceutical manufacturing industry takes only 10% of the market, and the export market for pharmaceuticals is insignificant despite the SADC providing a US$4.7 billion market. Inefficient production lines and high labour costs contribute significantly to high production costs, resulting in uncompetitive pricing for the locally manufactured pharmaceuticals against the imported ones. This research aims to enhance local pharmaceutical manufacturing machinery’s operational efficiency through machine learning (ML) to achieve competitiveness in local and regional markets. The operational efficiency targeted ML for real-time continuous inline quality control, reduced labour for machine operation, and overall enabling continuous manufacturing of pharmaceuticals to reduce throughput and waste in the form of in-process inventory and numerous start-up discards. Literature review on the application of ML on pharmaceutical equipment reviewed that several first world companies have adopted artificial intelligence (AI) and ML to improve their operations. Case research on PharmaCo revealed that labour and material usage during tabletting was the significant production cost drivers. An algorithm was developed using Python programing language, and it utilised the linear relationship between product properties, machine parameters, and the product output. The ML algorithm was used for simulating the impact of ML on the operational efficiency of the local pharmaceutical machinery. The population was the local pharmaceuticals manufacturing companies. ML technology was simulated at different manufacturing stages. The ML adoption was the only variable in the research as other product costing variables were either optimised or constant. Through simulation, the ML algorithm proved to be both effective and efficient in controlling a tablet-making machine. When varying flow rate, data was introduced for predicting corresponding machine speed to give tablets matching the intent weight, with at least 99.8% confidence, as calculated from the linear model. Other than the enhanced quality benefits, the ML algorithm enabled a remarkable reduction of labour by 50%, material waste by up to 4% (about 80% of initial waste), and set up time by 13%. Adopting ML thus enables a reduction of total production costs by up to 19%, which translates to a 19% reduction in the price of the locally manufactured tablets. Using the Ibuprofen 400 mg tablets as an example that are 10% pricier than the imported ones, it was noted that indeed ML could improve pharmaceutical machinery and make the local industry competitive enough to expand their market share locally regionally. By varying ML adoption levels and managing the costs of ML implementation, a break-even point was achieved. The minimum ML adoption level that would realise at least a 19% reduction in production cost was established.
