Reinforcement learning for inventory management in information-sharing pharmaceutical supply chains
| dc.contributor.advisor | Van Vuuren, J. H. | en_ZA |
| dc.contributor.advisor | Van Eeden, Joubert | en_ZA |
| dc.contributor.author | Du Plessis, Marno | en_ZA |
| dc.contributor.other | Stellenbosch University. Faculty of Industrial Engineering. Dept. of Industrial Engineering. | en_ZA |
| dc.date.accessioned | 2020-02-03T12:46:47Z | |
| dc.date.accessioned | 2020-04-28T12:13:30Z | |
| dc.date.available | 2020-02-03T12:46:47Z | |
| dc.date.available | 2020-04-28T12:13:30Z | |
| dc.date.issued | 2020-03 | |
| dc.description | Thesis (MEng)--Stellenbosch University, 2020. | en_ZA |
| dc.description.abstract | ENGLISH ABSTRACT: A general lack of information sharing across the various tiers of pharmaceutical supply chains in developing countries continues to compromise the availability of essential medicines. In the South African public health-care context, recent efforts aimed at improving information sharing in the pharmaceutical supply chain have been plagued by several implementation problems. It is conjectured that the true potential impact of information sharing remains unclear in the South African public health-care domain. The objective in this thesis is to elucidate conceptually how information sharing may benefit inventory management in a pharmaceutical supply chain. A number of hypothetical information-sharing scenarios are proposed in this thesis and their relative effectiveness is evaluated within a simulation modelling environment. The first of these scenarios does not involve any information sharing and serves as a benchmark. The scope of information sharing is further increased incrementally over the remaining scenarios. An agentbased pharmaceutical supply chain simulation model is further established in this thesis in order to evaluate the impact of information sharing in the context of the information-sharing scenarios. This simulation model is implemented as a concept demonstrator and takes as input any userspecified supply chain network. The concept demonstrator is capable of modelling the high-level operation of a pharmaceutical supply chain over time, with a particular focus on the flow of inventory. A reinforcement learning approach is adopted towards discovering effective inventory replenishment policies, specifically informed by information sharing, during each of the aforementioned information-sharing scenarios. The effectiveness of these policies is measured in respect of the total number of stock-outs and product expiries observed in the supply chain. A comparative analysis of the information-sharing scenarios is performed in the context of a hypothetical supply chain network experiencing a fluctuating demand pattern. This analysis reveals that stock-outs may be mitigated substantially when allowing health-care facilities that are located in close proximity to one another to share inventory among each other. It is also shown that the types of, and granularity of, information shared are instrumental in determining the relative effectiveness of information sharing. | en_ZA |
| dc.description.abstract | AFRIKAANSE OPSOMMING: 'n Algemene gebrek aan die deel van inligting oor verskillende vlakke van farmaseutiese voorsieningskettings in ontwikkelende lande belemmer die beskikbaarheid van noodsaaklike medisyne. In die Suid-Afrikaanse openbare gesondheidsorg-konteks is onlangse pogings om die deel van inligting in die farmaseutiese voorsieningsketting te verbeter, geteister deur verskeie implementeringsprobleme. Daar word vermoed dat die werklike potensiele impak van inligting-deling in Suid-Afrikaanse openbare gesondheidsorg onduidelik is. Die doel van hierdie tesis is om konseptueel toe te lig hoe die deel van inligting voorraadbestuur in 'n farmaseutiese voorsieningsketting kan bevoordeel. 'n Aantal hipotetiese scenario's word vir die deel van inligting in hierdie tesis voorgestel en die relatiewe doeltreffendheid daarvan word in 'n simulasiemodelleringsomgewing beoordeel. Die eerste van hierdie scenario's behels geen inligting-deling nie en dien as maatstaf. Die omvang van inligting-deling word in die daaropvolgende scenario's inkrementeel verhoog. 'n Agentgebaseerde simulasiemodel vir farmaseutiese voorsieningskettings word verder in hierdie tesis daargestel om die impak van inligting-deling in die konteks van die bostaande scenario's te evalueer. Hierdie simulasiemodel word as 'n konsepdemonstrasie geimplementeer en neem 'n gebruikersgespesifiseerde voorsieningskettingnetwerk as toevoer. Die konsepdemonstrasiemodel is daartoe in staat om die hoevlak-werking van 'n farmaseutiese voorsieningsketting oor tyd te modelleer, met 'n spesifieke fokus op die vloei van voorraad. 'n Versterkingsleerbenadering word gevolg om vir elk van die bogenoemde scenario's doeltreffende voorraadaanvullingsbeleide te ontdek deur spesifiek gebruik te maak van inligting-deling. Die doeltreffendheid van hierdie beleide word gemeet in terme van die totale aantal voorraadtekorte en produkverstrykings wat in die voorsieningsketting waargeneem word. 'n Vergelykende studie van die inligting-delingscenario's word in die konteks van 'n hipotetiese voorsieningskettingnetwerk met 'n wisselende vraagpatroon uitgevoer. Uit hierdie ontleding volg dit dat voorraadtekorte aansienlik verlaag kan word indien nabygelee gesondheidsorgfasiliteite voorraad onder mekaar kan uitruil. Daar word ook aangetoon dat die tipes inligting en die mate van inligting-deling instrumenteel is in die bepaling van die relatiewe doeltreffendheid van die deel van inligting. | af_ZA |
| dc.description.version | Masters | en_ZA |
| dc.format.extent | xxii, 176 leaves : illustrations | |
| dc.identifier.uri | http://hdl.handle.net/10019.1/107996 | |
| dc.language.iso | en | en_ZA |
| dc.publisher | Stellenbosch : Stellenbosch University | en_ZA |
| dc.rights.holder | Stellenbosch University | en_ZA |
| dc.subject | Reinforcement learning | en_ZA |
| dc.subject | Information sharing | en_ZA |
| dc.subject | Healthcare -- Inventory control -- South Africa | en_ZA |
| dc.subject | Supply chain management | en_ZA |
| dc.subject | Medical supplies -- Purchasing | en_ZA |
| dc.subject | Pharmaceutical supply chain | en_ZA |
| dc.subject | UCTD | en_ZA |
| dc.title | Reinforcement learning for inventory management in information-sharing pharmaceutical supply chains | en_ZA |
| dc.type | Thesis | en_ZA |