Academic literature on the topic 'Uncertainty treatment'

In 2017, 1.6 million people worldwide died from tuberculosis (TB). A new TB diagnostic test—Xpert MTB/RIF from Cepheid—was endorsed by the World Health Organization in 2010. Trials demonstrated that Xpert is faster and has greater sensitivity and specificity than smear microscopy—the most common sputum-based diagnostic test. However, subsequent trials found no impact of introducing Xpert on morbidity and mortality. We present a decision-theoretic model of how a clinician might decide whether to order Xpert or other tests for TB, and whether to treat a patient, with or without test results. Our first result characterizes the conditions under which it is optimal to perform empirical treatment; that is, treatment without diagnostic testing. We then examine the implications for decision making of partial knowledge of TB prevalence or test accuracy. This partial knowledge generates ambiguity, also known as deep uncertainty, about the best testing and treatment policy. In the presence of such ambiguity, we show the usefulness of diversification of testing and treatment.

Current treatments of uncertainty in environmental research embody several myths about the causes and consequences of imperfect knowledge, namely: (1) the dominant role of environmental factors in controlling uncertainty, such as system complexity, non-linearity and space-time variability, rather than social and psychological factors; (2) the primacy of observations in locating, quantifying, and reducing uncertainty; and (3) the value of technical assessments of uncertainty in ‘risk-based decision-making’. While the identification and treatment of specific sources of uncertainty remain impractical in some areas of environmental research, a source-based approach is increasingly used in environmental modeling. Here, selected sources of uncertainty are quantified with probability distributions and propagated to model outputs (a forward problem), while data are used to calibrate these estimates and reduce uncertainty (an inverse problem). More generally, current treatments of uncertainty and risk are dominated by attempts to quantify, minimize, and control uncertainty. Uncertainty is viewed as an ‘information deficit’ to be resolved, rather than an inherent product of conducting research. This paper argues for more open treatments of uncertainty in environmental research. Such openness requires an appreciation of the social and psychological causes of uncertainty, the role of observations as imperfect and contingent expressions of visible events, and the myriad ways in which scientific information can be misinterpreted, misused, or sidelined in environmental decision-making. The paper begins with a discussion of the nature and causes of uncertainty in environmental research. A review of current treatments of uncertainty is followed by an analysis of the source-based approach to assessing uncertainty. Prospects for the open treatment of uncertainty are then discussed in terms of circumventing the three ‘myths of uncertainty’ that characterize recent work in environmental research.

The inputs to probabilistic seismic hazard studies (seismic source zones, earthquake rates, attenuation functions, etc.) are uncertain, being based on subjective judgments and interpretations of limited data. In the face of this uncertainty, we consider (a) how one might “reasonably” determine the ground-motion levels to show on a single probabilistic seismic hazard map, and (b) the extent to which uncertainty in the calculated levels can be meaningfully represented on such a map. If the “best guess” estimates of the earthquake rate, the Gutenberg-Richter b-value and the maximum magnitude for a single source zone are regarded as uncorrelated and the uncertainty in each parameter can be regarded as symmetric about the estimated value, then the probabilistic ground-motion levels calculated using these best estimates represent both most likely values and also approximate mean values.

Classifiers are often tested on relatively small data sets, which should lead to uncertain performance metrics. Nevertheless, these metrics are usually taken at face value. We present an approach to quantify the uncertainty of classification performance metrics, based on a probability model of the confusion matrix. Application of our approach to classifiers from the scientific literature and a classification competition shows that uncertainties can be surprisingly large and limit performance evaluation. In fact, some published classifiers may be misleading. The application of our approach is simple and requires only the confusion matrix. It is agnostic of the underlying classifier. Our method can also be used for the estimation of sample sizes that achieve a desired precision of a performance metric.

This paper serves as a problem statement of the issues surrounding uncertainty in wastewater treatment modelling. The paper proposes a structure for identifying the sources of uncertainty introduced during each step of an engineering project concerned with model-based design or optimisation of a wastewater treatment system. It briefly references the methods currently used to evaluate prediction accuracy and uncertainty and discusses the relevance of uncertainty evaluations in model applications. The paper aims to raise awareness and initiate a comprehensive discussion among professionals on model prediction accuracy and uncertainty issues. It also aims to identify future research needs. Ultimately the goal of such a discussion would be to generate transparent and objective methods of explicitly evaluating the reliability of model results, before they are implemented in an engineering decision-making context.

This dissertation addresses the issue of robustness in proton therapy treatment planning for cancer treatment. Proton therapy is considered to be advantageous in treating most childhood cancers and certain adult cancers, including those of the skull base, spine and head and neck. Protons, unlike X-rays, have a finite range highly dependent on the electron density of the material they are traversing, resulting in a steep dose gradient at the distal edge of the Bragg peak. These characteristics, together with advancements in computation and technology have led to the ability to plan and deliver treatments with greater conformality, sparing normal tissue and organs at risk. Radiotherapy treatment plans aim to meet set dosimetric constraints, and meet them at every fraction. Plan robustness is a measure of deviation between the delivered dose distribution and the planned dose distribution. Due to the same characteristics that make protons advantageous, conventional means of using margins to create a Planning Target Volume (PTV) to ensure plan robustness are inadequate. Additional to this, without a PTV, a new method of analysing plan quality is required in proton therapy. My original contribution to the knowledge in this area is the demonstration of how site- and centre- specific robustness constraints can be established. Robustness constraints can be used both for proton plan analysis and to identify patients that require plans of greater individualisation. I have also used the daily volumetric imaging from patients previously treated with conventional radiotherapy to quantify range uncertainty from inter- and intra-fraction motion. These new methods of both quantifying and analysing the change in proton range in the patient can aid in the choice of beam directions, provide input into a multi- criteria optimisation algorithm or can be used as criteria to determine when adaptive planning may be required. This greater understanding in range uncertainty better informs the planner on how best to balance the trade-off between plan conformality and robustness in proton therapy. This research is directly relevant to furthering the knowledge base in light of HM Government pledging £250 million to build two proton centres in England, to treat NHS patients from 2018. Use of methods described in this dissertation will aid in the establishment of clear and pre-defined protocols for treating patients in the future.

Bibliography: leaves 142-149.
The nature of human decision making dictates that a decision must often be considered under conditions of uncertainty. Decisions may be influenced by uncertain future events, doubts regarding the precision of inputs, doubts as to what the decision maker considers important, and many other forms of uncertainty. The multicriteria decision models that are designed to facilitate and aid decision making must therefore consider these uncertainties if they are to be effective. In this thesis, we consider the treatment of uncertainty in multicriteria decision making (MCDM), with a specific view to investigating the types of uncertainty that are most relevant to MCDM, [and] how the uncertainties identified as relevant may be treated by various different MCDM methodologies.

The purpose of this thesis was to acquaint the reader on the nature of the uncertainty present in construction price forecasting and to introduce an environment that has the ability to integrate these uncertainties with greater consistency than that possessed by available price models. The objective of this thesis was to establish that uncertainty can be explicitly treated in price forecasting models. This would have two benefits to concerned parties. Firstly, the effectiveness of price forecasts could be improved as provision could be made for any uncertain variable. This will be of great benefit to the client, as a more accurate assessment of the building process could be established at an earlier stage of the design process. Secondly, the price forecast will be more useful to quantity surveyors, architects and clients, as it would 'quantify' the extent of the uncertainty which could be provided for in a more meaningful manner. In order to establish that existing price models do not deal with the uncertainty present at the time of forecasting, the price models used by practitioners were evaluated against the different types of uncertainty found at the different stages of the price forecasting process. Once this had been established, eight techniques that have the ability to treat various forms of uncertainty, were presented. After analysing the techniques abilities to cope with the uncertainties associated with price forecasting, it was established that certain of these techniques do have the ability, and are suitable, to be incorporated into the price forecasting process. From the results of a questionnaire survey conducted on quantity surveying offices in South Africa, it was found that the price models used by practitioners do not take uncertainty into account, and have in fact, the potential for uncertainty inducement. Some of the uncertainty found to be present in the preparation of a construction price forecast include the lack or incompleteness of design information, the uncertainty in the communication of design information, the variability in the data used by quantity surveyors and, the uncertainty in the choice of price model during the different stages of the design process. As a possible solution to the problem of uncertainty, an expert system environment, utilising a three-dimensional classification of uncertainty, has been proposed. It has been proved that this environment has the ability to cater for the uncertainty associated with the price forecasting process, as well as having the attribute of providing the user with the reasoning behind the logic that the expert system has followed, a characteristic not possible with the traditional forms of price models. From the findings of this thesis, it can be concluded that the methods of price modelling used by quantity surveying practitioners, are unable to take uncertainty into account effectively. It can also be concluded that an expert system environment has the ability to handle the different forms of uncertainty found at the various stages of construction design. The proposed model is conceptual in nature and has not been tested in practice. It is therefore recommended that further research be carried out in this field, with the aim of producing a construction price forecasting expert system which utilises the proposed three-dimensional classification of uncertainty.

2010/2011
Nello studio dei sistemi di trasporto l’acquisizione e l’utilizzo di informazioni corrette e aggiornate sullo stato dei sistemi rappresentano da sempre elementi di centrale importanza per la produzione di analisi adeguate ed affidabili. Sfortunatamente in molti ambiti applicativi le informazioni disponibili per le analisi sono invece spesso carenti o di bassa qualità, e il loro utilizzo si traduce in risultati affetti da elevata incertezza e talvolta di dubbia validità. I processi di evoluzione tecnologica che interessano campi quali l’informatica, l’elettronica e le telecomunicazioni stanno rendendo progressivamente più semplice e conveniente l’acquisizione di rilevanti quantità di dati di interesse per le analisi trasportistiche, sia tradizionalmente raccolti per studi trasportistici (ad esempio dati di traffico rilevati su tronchi stradali) sia non direttamente connessi ad un uso trasportistico (ad esempio segnali Bluetooth e GPS provenienti da dispositivi di telefonia mobile). Tuttavia in molti casi l’ampia disponibilità di dati, soprattutto nel secondo caso, non si traduce in immediata spendibilità applicativa. I dati sono infatti spesso disomogenei dal punto di vista informativo, caratterizzati da una qualità non necessariamente elevata e spesso richiedono onerosi processi di verifica e validazione. In questi particolari contesti l’applicazione di tecniche di Data Mining può dimostrarsi una soluzione indubbiamente vantaggiosa. Esse infatti, per loro intrinseca natura, rendono possibile la gestione efficace di grandi quantità di dati e la produzione di risultati sempre più robusti all’aumentare delle dimensioni della base di dati disponibile. Sulla base di queste considerazioni questo lavoro di tesi si è concentrato in primo luogo su un’attenta revisione delle più consolidate tecniche di Data Mining, individuando gli ambiti applicativi, nel campo dei trasporti, in cui esse possono rappresentare dei validi strumenti di analisi. Con il termine Data Mining si fa riferimento al processo di estrazione dell’informazione presente in un certo insieme di dati, finalizzato ad individuare relazioni “nascoste” nei dati stessi o comunque a sintetizzare in modalità nuove la visione su di essi. Esso rappresenta una parte di un più ampio processo di estrazione della conoscenza, che inizia con un’accurata selezione e trasformazione dei dati disponibili (come detto i dati sottoposti a “mining” sono spesso raccolti con altri obiettivi) e si conclude con un’attenta interpretazione e valutazione dei risultati. Uno schema di classificazione generalmente accettato suddivide le tecniche di Data Mining in sei categorie in rapporto alla funzione considerata: stima (reti neurali, modelli di regressione, alberi decisionali), previsione (reti neurali, alberi decisionali), classificazione (k-nearest neighbour, alberi decisionali, reti neurali), raggruppamento (tecniche di clustering, Self-Organising-Maps), associazione (regole di associazione) e descrizione (regole di associazione, clustering, alberi decisionali). Nel presentare un quadro d’insieme dell’ampia letteratura esistente in materia, uno specifico rilievo è stato dato alle più consolidate tecniche di classificazione, raggruppamento e associazione, in quanto maggiormente impiegate nei diversi contesti applicativi. Successivamente è stato tracciato uno stato dell’arte per ciò che attiene le applicazioni in ambito trasportistico. In tal senso la revisione dei lavori prodotti ha evidenziato la notevole flessibilità d’uso di queste tecniche e la loro crescente diffusione applicativa. Molti sono infatti i filoni di ricerca che hanno beneficiato di queste tecniche innovative; tra questi nel lavoro di tesi si sono evidenziati alcuni tra i più interessanti: la previsione a breve termine dei flussi di traffico da dati storici o in real-time (traffic forecasting), l’identificazione e la quantificazione dei fattori che influenzano i fenomeni di incidentalità, l’analisi di sistemi di gestione delle pavimentazioni stradali e di sistemi di monitoraggio del traffico. La seconda parte della tesi si è invece focalizzata su un’applicazione delle tecniche di Data Mining allo studio del funzionamento di un sistema viario, attraverso una revisione critica della Procedura FHWA (Federal Highway Administration) per il monitoraggio del traffico stradale. La scelta di questo filone di ricerca è data dal fatto che la raccolta di informazioni sui volumi di traffico è un aspetto rilevante nell’attività di pianificazione dei trasporti (ambito stradale), quale componente significativa del processo conoscitivo. D’altra parte i costi legati alla gestione dei sistemi di monitoraggio, sia per attrezzature che per personale, richiedono una crescente attenzione alla loro progettazione, al fine di ottenere la massima qualità dei risultati. Negli Stati Uniti la FHWA definisce periodicamente alcune linee guida per migliorare questi aspetti attraverso la Traffic Monitoring Guide (2001) e ha raggiunto progressivamente un ruolo di riferimento per altre agenzie dello stesso tipo in altre parti del mondo, Italia compresa. Tale procedura è basata sull’uso congiunto di rilievi di diversa durata (rilievi in continuo con strumenti fissi e rilievi di breve durata con apparecchiature portatili) ed è finalizzata principalmente alla stima del Traffico Giornaliero Medio Annuo (Annual Average Daily Traffic, AADT). L’analisi della letteratura esistente ha individuato la lacuna principale della procedura FHWA nella determinazione dei gruppi tipologici di strade sulla base dei profili temporali di traffico e nell’assegnazione delle sezioni monitorate con rilievi di breve durata a questi gruppi. L’approccio elaborato si è pertanto posto l’obiettivo di migliorare la procedura relativamente a questi due aspetti rilevanti. Per trattare l’esistenza di situazioni di incerta attribuzione di una sezione stradale ad un certo gruppo tipologico, specie quando non è semplice fornire una chiara definizione in termini trasportistici (ad esempio strada “pendolare” o “turistica”), sono state adottate tecniche di Fuzzy Clustering, garantendo un’opportuna trattazione formale del problema. Per quanto concerne il secondo aspetto, le sezioni non monitorate in continuo vengono inserite nel gruppo tipologico più simile rispetto ai profili temporali di traffico osservati. Per effettuare l’assegnazione di queste sezioni ai gruppi tipologici, l’approccio proposto ha utilizzato una Rete Neurale Artificiale, opportunamente progettata per mantenere l’incertezza presente nella fase di creazione dei gruppi fino alla fine del processo. L’output della rete è infatti rappresentato dall’insieme delle probabilità di appartenenza del rilievo di breve durata ai diversi gruppi tipologici ed è interpretato utilizzando la teoria di Dempster-Shafer. Le misure di incertezza associate all’output (indici di non-specificità e discordanza) permettono di descrivere sinteticamente la qualità dell’informazione disponibile. L’approccio proposto è stato implementato considerando i dati di monitoraggio provenienti dal programma SITRA (Sistema Informativo TRAsporti) della Provincia di Venezia. Rispetto all’ambito applicativo di interesse è stata verificata la validità dell’approccio, confrontando i risultati ottenuti nella stima dell’AADT con precedenti approcci proposti in letteratura. L’analisi comparativa dei risultati ha permesso di rilevare una migliore accuratezza delle stime e soprattutto la possibilità, assente nei precedenti approcci, di evidenziare eventuali carenze informative (dovute all’esiguo numero di dati) e la necessità di procedere con ulteriori rilievi di traffico. I risultati positivi ottenuti in questa fase sperimentale hanno permesso di avviare il progetto per la realizzazione di uno strumento software di immediata spendibilità applicativa
In the study of transportation systems, the collection and the use of correct information of the state of the system represent a central point for the development of reliable and proper analyses. Unfortunately in many application fields information is generally obtained using limited, scarce and low-quality data and their use produces results affected by high uncertainty and in some cases low validity. Technological evolution processes which interest different fields, including Information Technology, electronics and telecommunications make easier and less expensive the collection of large amount of data which can be used in transportation analyses. These data include traditional information gathered in transportation studies (e.g. traffic volumes in a given road section) and new kind of data, not directly connected to transportation needs (i.e. Bluetooth and GPS data from mobile phones). However in many cases, in particular for the latter case, this large amount of data cannot be directly applied to transportation problems. Generally there are low-quality, non-homogeneous data, which need time consuming verification and validation process to be used. Data Mining techniques can represent an effective solution to treat data in these particular contexts since they are designed to manage large amount of data producing results whose quality increases as the amount of data increases. Based on these facts, this thesis first presents a review of the most well-established Data Mining techniques, identifying application contexts in transportation field for which they can represent useful analysis tools. Data mining can be defined as the process of exploration and analysis which aims to discover meaningful patterns and ‘’hidden’’ rules in the set of data under analysis. Data Mining could be considered a step of a more general Knowledge Discovery in Databases Process, which begins with selection, pre-processing and transformation of data (“mined” data are generally collected for reasons different from the analysis) and is completed with the interpretation and evaluation of results. A classification scheme generally accepted identifies six categories of DM techniques, which are related to the objective one would achieve from the analysis: estimation (neural networks, regression models, decision trees), prediction (neural networks, decision trees), classification (k-nearest neighbor, decision trees, neural networks), clustering (clustering techniques, Self-Organizing-Maps), affinity grouping or association (association rules) and profiling (association rules). In the review of the wide literature concerning Data Mining methods, particular attention has been devoted to the well-established technique of clustering, classification and association, since they are the most applied in different application contexts. The literature review process has been further extended to Data Mining applications in the transportation field. This review highlights the great flexibility of use of these techniques and the increasing number of applications. Many research topics have taken advantages of these innovative tools and some of them are presented due to their interest: short-term traffic flow forecasting from historical and real-time data, identification and quantification of factor risks in accident analysis, analysis of pavement management systems and traffic monitoring systems. The second part of the thesis has focused on the application of Data Mining techniques to road system analysis, through a critical review of U.S. Federal Highway Administration (FHWA) traffic monitoring approach. The choice of this topic is due to the fact that traffic monitoring activities represent a relevant aspect of highway planning activities, as a part of the knowledge process. However data collection activities produce relevant management costs, both for equipment and personnel, therefore monitoring programs need to be designed with attention to obtain the maximum quality of results. In the U.S.A., the Federal Highway Administration (FHWA) provides guidance for improving these aspects by way of its Traffic Monitoring Guide (TMG) (FHWA, 2001), which has a reference role for other similar agencies in the world. The FHWA procedure is based on two types of counts (short duration counts taken with portable traffic counters and continuous counts taken with fixed counters) and has the main objective of determine the Annual Average Daily Traffic (AADT). Critical review of literature on this topic has pointed out that the most critical aspects of this procedure are the definition of road groups based on traffic flow patterns and the assignment of a section to a road group using short counts. The proposed approach has been designed to solve both issues. The first issue is related to situations for which road section could belong to more than one road group, and the groups cannot be easily defined in transportation terms, (e.g. “commuter road”, “recreational road”). The proposed approach introduces Fuzzy Clustering techniques, which adopt an analytical framework consistent with this kind of uncertainty. Concerning the second issue, road sections monitored with short counts are assigned to the road group with more similar traffic patterns. In the proposed approach an Artificial Neural Network is implemented to assign short counts to roads groups. The Neural network is specifically designed to maintain the uncertainty related to the definition of road groups until the end of the estimation process. In fact the output of the Neural Network are the probabilities that the a specific short counts belongs to the road groups. These probabilities are interpreted using the Dempster-Shafer theory; measures of uncertainty related to the output (indices of non-specificity and discord) provide an assessment of the quality of information in a synthetic manner. The proposed approach have been implement on a case study, using traffic data from SITRA (Sistema Informativo TRAsporti) monitoring program of the Province of Venice. In this specific context the approach has been validated and the results obtained (AADT estimates) from the proposed method have been compared with those obtained by two approaches proposed in previous studies. The comparative analysis highlights that the proposed approach increases the accuracy of estimates and gives indication of the quality of assignment (depending on sample size) and suggests the need for additional data collection. The positive results obtained in the experimental phase of the research have led to the design of a software tool to be used in next future in real world applications.
XXIV Ciclo
1983

AbstractThis chapter presents the technique route of model updating in the presence of imprecise probabilities. The emphasis is put on the inevitable uncertainties, in both numerical simulations and experimental measurements, leading the updating methodology to be significantly extended from deterministic sense to stochastic sense. This extension requires that the model parameters are not regarded as unknown-but-fixed values, but random variables with uncertain distributions, i.e. the imprecise probabilities. The final objective of stochastic model updating is no longer a single model prediction with maximal fidelity to a single experiment, but rather the calibrated distribution coefficients allowing the model predictions to fit with the experimental measurements in a probabilistic point of view. The involvement of uncertainty within a Bayesian updating framework is achieved by developing a novel uncertainty quantification metric, i.e. the Bhattacharyya distance, instead of the typical Euclidian distance. The overall approach is demonstrated by solving the model updating sub-problem of the NASA uncertainty quantification challenge. The demonstration provides a clear comparison between performances of the Euclidian distance and the Bhattacharyya distance, and thus promotes a better understanding of the principle of stochastic model updating, as no longer to determine the unknown-but-fixed parameters, but rather to reduce the uncertainty bounds of the model prediction and meanwhile to guarantee the existing experimental data to be still enveloped within the updated uncertainty space.

The authors have studied daily decision-making processes in groups under uncertainty, with an exploratory field study in the medical domain. The work follows the tradition of naturalistic decision-making (NDM) research. It aims to understand how groups in this high reliability context conceptualize and internalize uncertainties, and how they handle them in order to achieve effective decision-making in their everyday activities. Analysis of the survey data shows that uncertainty is thought of in terms of issues and sources (as identified by previous research), but also (possibly a domain-specific observation) as a lack of personal knowledge or skill. Uncertainty is accompanied by emotions of fear and shame. It arises during the diagnostic process, the treatment process and the outcome of medical decision making. The most frequently cited sources of uncertainty are partly lacking information and inadequate understanding owing to instability of information. Descriptions of typical group decisions reveal that the individual himself is a source of uncertainty when a lack of knowledge, skills and expertise is perceived. The group can serve as a source of uncertainty if divergent opinions in the decision making group exist. Three different situations of group decisions are identified: Interdisciplinary regular meetings (e.g. tumor conferences), formal ward meetings and ad hoc consultations. In all healthcare units concerned by the study, only little use of structured decision making procedures and processes is reported. Strategies used to handle uncertainty include attempts to reduce uncertainty by collecting additional information, delaying action until more information is available or by soliciting advice from other physicians. The factors which ultimately determine group decisions are hierarchy (the opinion of more senior medical staff carries more weight than that of junior staff), patients’ interest and professional competence. Important attributes of poor group decisions are the absence of consensus and the use of hierarchy as the predominant decision criterion. On the other hand, decisions judged to be effective are marked by a sufficient information base, a positive discussion culture and consensus. The authors identify four possible obstacles to effective decision making: a steep hierarchy gradient, a poor discussion culture, a strong need for consensus, and insufficient structure and guidance of group decision making processes. A number of intervention techniques which have been shown in other industries to be effective in improving some of these obstacles are presented.

Addressing the public health threat posed by AMR is a national strategic priority for the UK, which has led to both a 20-year vision of AMR and a 5-year (2019 to 2024) AMR National Action Plan (NAP). The latter sets out actions to slow the development and spread of AMR with a focus on antimicrobials. The NAP used an integrated ‘One-Health’ approach which spanned people, animals, agriculture and the environment, and calls for activities to “identify and assess the sources, pathways, and exposure risks” of AMR. The FSA continues to contribute to delivery of the NAP in a number of ways, including through furthering our understanding of the role of the food chain and AMR.Thorough cooking of food kills vegetative bacterial cells including pathogens and is therefore a crucial step in reducing the risk of most forms of food poisoning. Currently, there is uncertainty around whether cooking food is sufficient to denature AMR genes and mobile genetic elements from these ‘dead’ bacteria to prevent uptake by ‘live’ bacteria in the human gut and other food environments - therefore potentially contributing to the overall transmission of AMR to humans. This work was carried out to assess these evidence gaps.

Taxpayers sometimes engage in complex transactions with uncertain tax treatment, such as mergers, acquisitions, demergers and spin-offs. With the rise of global value chains and proliferation of multinational corporations, these transactions increasingly involve transnational financial arrangements and cross-border dealings, making tax treatment even more uncertain. If improperly structured, such transactions could have costly tax consequences. One approach to dealing with this uncertainty is to create a private rulings regime, whereby a taxpayer applies for a private ruling by submitting a statement detailing the transaction (proposed or completed) to the tax authority. The tax authority interprets and applies the tax laws to the requesting taxpayer’s specific set of facts in a written private ruling. The private ruling offers taxpayers certainty as to how the tax authority views the transaction, and the tax treatment the taxpayer can expect based on the specific facts presented. Private rulings are a common feature of many tax systems around the world, and their main goal is to promote tax certainty and increase investor confidence in the tax system. This is especially important in a developing country like Uganda, whose tax laws are often amended and may not anticipate emerging transnational tax issues. Private rulings in Uganda may be applied for in writing prior to or after engaging in the transaction. The Tax Procedures Code Act (TPCA), which provides for private rulings, requires applicants to make a full and true disclosure of the transaction before a private ruling may be issued. This paper evaluates the Ugandan private rulings regime, offering a comparative perspective by highlighting similarities and contrasts between the Ugandan regime and that of other jurisdictions, including the United States, Australia, South Africa and Kenya. The Ugandan private rulings regime has a number of strengths. It is not just an administrative measure as in some jurisdictions, but is based on statute. Rulings are issued from a central office – instead of different district offices, which may result in conflicting rulings. Rather than an elaborate appeals process, the private ruling is only binding on the URA and not on the taxpayer, so a dissatisfied taxpayer can simply ignore the ruling. The URA team that handles private rulings has diverse professional backgrounds, which allows for a better understanding of applications. There are, however, a number of limitations of the Ugandan private rulings system. The procedure of revocation of a private ruling is uncertain. Private rulings are not published, which makes them a form of ‘secret law’. There is no fee for private rulings, which contributes to a delay in the process of issuing one. There is understaffing in the unit that handles private rulings. Finally, there remains a very high risk of bias against the taxpayer because the unit is answerable to a Commissioner whose chief mandate is collection of revenue. A reform of the private rulings regime is therefore necessary, and this would include clarifying the circumstances under which revocation may occur, introducing an application fee, increasing the staffing of the unit responsible, and placing the unit under a Commissioner who does not have a collection mandate. While the private rulings regime in Uganda has shortcomings, it remains an essential tool in supporting investor confidence in the tax regime.

Review question / Objective: It remains uncertain whether first-line treatment with upfront brain radiotherapy (RT) in combination with epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) is superior to EGFR-TKIs alone in EGFR-mutated non-small-cell lung cancer with newly diagnosed brain metastases (BMs). We performed a meta-analysis to address this issue. Condition being studied: Brain radiotherapy (RT) has been shown to damage the blood-brain barrier (BBB) and improve the concentration of EGFR-TKIs in the CSF. Additionally, RT can result in a reduction of EGFR-TKIs resistance. Therefore, EGFR-TKIs in combination with brain RT should be more effective than EGFR-TKIs alone theoretically. However, results from retrospective studies are inconsistent. There is the possibility that patients characteristics or brain RT technique affect the efficacy of treatments. To date, there is still no randomized controlled trials (RCTs) comparing the two treatment strategies.

Tax amnesties have taken centre stage as a compliance tool in recent years. The OECD estimates that since 2009 tax amnesties in 40 jurisdictions have resulted in the collection of an additional €102 billion in tax revenue. A number of African countries have introduced tax amnesties in the last decade, including Nigeria, Namibia, South Africa and Tanzania. Despite their global popularity, the efficacy of tax amnesties as a tax compliance tool remains in doubt. The revenue is often below expectations, and it probably could have been raised through effective use of regular enforcement measures. It is also argued that tax amnesties might incentivise non-compliance – taxpayers may engage in non-compliance in the hope of benefiting from an amnesty. This paper examines the administration of tax amnesties in various jurisdictions around the world, including the United States, Australia, Canada, Kenya and South Africa. The paper makes a cost-benefit analysis of these and other tax amnesties – and from this analysis develops a model tax amnesty, whose features maximise the benefits of a tax amnesty while minimising the potential costs. The model tax amnesty: (1) is permanent, (2) is available only to taxpayers who make a voluntary disclosure, (3) relieves taxpayers of penalties, interest and the risk of prosecution, but treats intentional and unintentional non-compliance differently, (4) has clear reporting requirements for taxpayers, and (5) is communicated clearly to attract non-compliant taxpayers without appearing unfair to the compliant ones. The paper then focuses on the Ugandan tax amnesty introduced in July 2019 – a Voluntary Disclosure Programme (VDP). As at 7 November 2020, this initiative had raised USh16.8 billion (US$6.2 million) against a projection of USh45 billion (US$16.6 million). The paper examines the legal regime and administration of this VDP, scoring it against the model tax amnesty. It notes that, while the Ugandan VDP partially matches up to the model tax amnesty, because it is permanent, restricted to taxpayers who make voluntary disclosure and relieves penalties and interest only, it still falls short due to a number of limitations. These include: (1) communication of the administration of the VDP through a public notice, instead of a practice note that is binding on the tax authority; (2) uncertainty regarding situations where a VDP application is made while the tax authority has been doing a secret investigation into the taxpayer’s affairs; (3) the absence of differentiated treatment between taxpayers involved in intentional non-compliance, and those whose non-compliance may be unintentional; (4) lack of clarity on how the VDP protects the taxpayer when non-compliance involves the breach of other non-tax statutes, such as those governing financial regulation; (5)absence of clear timelines in the administration of the VDP, which creates uncertainty;(6)failure to cater for voluntary disclosures with minor errors; (7) lack of clarity on VDP applications that result in a refund position for the applicant; and (8) lack of clarity on how often a VDP application can be made. The paper offers recommendations on how the Ugandan VDP can be aligned to match the model tax amnesty, in order to gain the most from this compliance tool.