Refine
Document Type
- Study Thesis (2)
- Article (1)
- Bachelor Thesis (1)
Language
- English (4)
Has Fulltext
- yes (4)
Keywords
- 1D-CNN (1)
- Botswana (1)
- Deep learning (1)
- Digitization (1)
- Maun Science Park (2)
- Obstructive sleep apnea (1)
- Portable monitor (1)
- Sleep Apnea (1)
- Smart Farming (1)
- Vertical Farming (1)
Institute
Global agriculture will face major challenges in the future. In addition to the increasing demand for food due to constant population growth, the consequences of climate change will make it even more difficult to operate agriculture and supply people with food. In addition to further productivity increases in traditional agriculture, new concepts for sustainable and scalable food production are needed. Vertical farming offers a promising approach.
The aim of this project is to demonstrate how vertical farming can be used to ensure sustainable food production and how this concept can be applied in the pioneering Maun Science Park project in Botswana. In doing so, the Maun Science Park will address future challenges such as demographics, governance and climate change and become a best practice model for Botswana, the whole of Africa and the world. The country of Botswana grew to become one of the most prosperous countries in Africa in recent decades due to strong economic growth from mining. However, the population faces great challenges in the future; in addition to great social inequality, climate change threatens the country's overall supply.
With the help of a literature review and qualitative and quantitative interviews with stakeholders from Maun (Botswana), the potentials and challenges for vertical farming in Botswana could be identified and future measures for a possible realization could be derived. Basically, some challenges in Botswana are addressed by the technology, for example, Vertical Farming offers high food security through year-round production of food through the closed ecosystem and creates independence from current and future climatic conditions, poor conditions for traditional agriculture (e.g. infertile soils) and foreign imports. However, the main structural problems of agriculture in Botswana, such as the lack of infrastructure, know-how and policy support, are not addressed.
Botswana serves as a role model for other African countries due to its rapid development in recent decades. Since the country is sparsely populated and a large part of the rural population depends on agriculture, especially livestock, this sector forms the backbone of the national economy. The digitization of this sector offers promising opportunities for economic growth and driving Botswana's evolution to a digital economy, while real value is being created for smallholder farmers. To support this process, an ITU research project made the key recommendation for the development of a digital crowdfarming tool and marketplace to create a digital ecosystem for smallholder agriculture. Within the research project, infrastructural challenges such as the creation of rural electricity supply and internet access, as well as the smallholders' need for remote monitoring, management, and better connectivity, were identified.
Based on the findings of the ITU research report, this bachelor's thesis aims to identify potential innovations for the digital development of smallholder agriculture in Botswana and to conceptualize proposals to address the identified challenges and needs of smallholder farmers. To achieve this, solutions were developed through literature research, technology analysis and expert involvement. These included the design of a decentralized mini-grid for power supply, proposals to create internet access, and the graphic visualization of a conceptual app. The latter addresses smallholder farmers' needs for remote monitoring, market access, knowledge enhancement, and connection to colleagues, buyers, and investors.
The proposed solutions and developed concepts provide impulses for further research and can serve as a basis for an extended evaluation through further involvement of experts and stakeholders.
Background: Polysomnography (PSG) is the gold standard for detecting obstructive sleep apnea (OSA). However, this technique has many disadvantages when using it outside the hospital or for daily use. Portable monitors (PMs) aim to streamline the OSA detection process through deep learning (DL).
Materials and methods: We studied how to detect OSA events and calculate the apnea-hypopnea index (AHI) by using deep learning models that aim to be implemented on PMs. Several deep learning models are presented after being trained on polysomnography data from the National Sleep Research Resource (NSRR) repository. The best hyperparameters for the DL architecture are presented. In addition, emphasis is focused on model explainability techniques, concretely on Gradient-weighted Class Activation Mapping (Grad-CAM).
Results: The results for the best DL model are presented and analyzed. The interpretability of the DL model is also analyzed by studying the regions of the signals that are most relevant for the model to make the decision. The model that yields the best result is a one-dimensional convolutional neural network (1D-CNN) with 84.3% accuracy.
Conclusion: The use of PMs using machine learning techniques for detecting OSA events still has a long way to go. However, our method for developing explainable DL models demonstrates that PMs appear to be a promising alternative to PSG in the future for the detection of obstructive apnea events and the automatic calculation of AHI.
