Andrew Conkie - Academia.edu (original) (raw)
Papers by Andrew Conkie
Heart failure, May 27, 2024
was not significant (p for interaction 0.52). All-cause first hospitalisation was reduced for tho... more was not significant (p for interaction 0.52). All-cause first hospitalisation was reduced for those with IHD assigned to FDI (HR 0.78, 95% CI 0.64-0.94) but not for those with no-IHD (HR 1.08, 95% CI 0.86-1.35; p for interaction 0.030). For patients with IHD and TSAT<20% (n = 471), there were fewer primary endpoint events, CV deaths and all-cause mortality but these differences did not reach significance (table 2). Conclusion For patients with HFrEF in the IRONMAN trial, FDI is associated with a trend to greater benefit in those with IHD, consistent with similar (non-significant) trends observed in the IPD meta-analysis. Acknowledgements The study was funded by the British Heart Foundation (grant award CS/15/1/31175). Pharmacosmos provided supplies of intravenous ferric derisomaltose and additional trial support with an unrestricted grant. We thank the participants, and all the staff who contributed to the IRON-MAN trial.
Zeitschrift f�r Wahrscheinlichkeitstheorie und Verwandte Gebiete, 1977
The purpose of the present paper is to extend the results of [2] from the continuous to the quasi... more The purpose of the present paper is to extend the results of [2] from the continuous to the quasi-left continuous case. We are indebted to P.-A. Meyer for pointing out that some of the results can be further extended to apply to regular processes. Our main goal is to show that any sequence {T(n)} of stopping times which does not grow too rapidly admits of a subsequence converging, as far as the process in question is concerned, to a finite stopping time T. The growth condition is just that limP{T(n)>a} =0 uniformly in n. In a~co order to make the result valid it is necessary to enlarge the probability space so that there is a random variable with uniform distribution on [0, 11 which is independent of the process and also to properly define the notion of convergence. The proper notion of convergence is easy to state and it is that the subsequence { T(n(k))} converges to T as far as the process is concerned provided that the distributions of {T(n(k))} and of the process stopped at T(n(k)) converge in the usual way, i.e. weakly, to the distributions of T and of the process stopped at T. The enlargement of the space which is required is essential since it is easy to give examples which show that the result is false without it. This enlargement gives rise to randomized stopping times, so that the closure of the space of stopping times is the spaces of randomized stopping times. A result of independent interest obtained in the course of the proof is that if {T(n)} and {U(n)} are sequences of stopping times which do not grow too rapidly in the sense already given and if T(n)-U(n) converges to zero in probability then f(Xr(n), Xv(,)) converges to zero in probability as well, where X~ is the process and where f(x, y) is a function which can be taken to be the distance between x and y, when one exists, although in general the state space is not assumed to be a metric space. An overall assumption made about the process is that it is standard, where the definition of the standard process is given in the same way as in the Markov case. It is not necessary, however, to assume that the process is a Markov process. Many probabilistic constructions involve a passage from the discrete to the continuous. The main theorem in the paper gives a general method for proving
Frontiers in Digital Health, 2022
Identifying which patients are at higher risks of dying or being re-admitted often happens to be ... more Identifying which patients are at higher risks of dying or being re-admitted often happens to be resource- and life- saving, thus is a very important and challenging task for healthcare text analytics. While many successful approaches exist to predict such clinical events based on categorical and numerical variables, a large amount of health records exists in the format of raw text such as clinical notes or discharge summaries. However, the text-analytics models applied to free-form natural language found in those notes are lagging behind the break-throughs happening in the other domains and remain to be primarily based on older bag-of-words technologies. As a result, they rarely reach the accuracy level acceptable for the clinicians. In spite of their success in other domains, the superiority of deep neural approaches over classical bags of words for this task has not yet been convincingly demonstrated. Also, while some successful experiments have been reported, the most recent bre...
Heart failure, May 27, 2024
was not significant (p for interaction 0.52). All-cause first hospitalisation was reduced for tho... more was not significant (p for interaction 0.52). All-cause first hospitalisation was reduced for those with IHD assigned to FDI (HR 0.78, 95% CI 0.64-0.94) but not for those with no-IHD (HR 1.08, 95% CI 0.86-1.35; p for interaction 0.030). For patients with IHD and TSAT<20% (n = 471), there were fewer primary endpoint events, CV deaths and all-cause mortality but these differences did not reach significance (table 2). Conclusion For patients with HFrEF in the IRONMAN trial, FDI is associated with a trend to greater benefit in those with IHD, consistent with similar (non-significant) trends observed in the IPD meta-analysis. Acknowledgements The study was funded by the British Heart Foundation (grant award CS/15/1/31175). Pharmacosmos provided supplies of intravenous ferric derisomaltose and additional trial support with an unrestricted grant. We thank the participants, and all the staff who contributed to the IRON-MAN trial.
Zeitschrift f�r Wahrscheinlichkeitstheorie und Verwandte Gebiete, 1977
The purpose of the present paper is to extend the results of [2] from the continuous to the quasi... more The purpose of the present paper is to extend the results of [2] from the continuous to the quasi-left continuous case. We are indebted to P.-A. Meyer for pointing out that some of the results can be further extended to apply to regular processes. Our main goal is to show that any sequence {T(n)} of stopping times which does not grow too rapidly admits of a subsequence converging, as far as the process in question is concerned, to a finite stopping time T. The growth condition is just that limP{T(n)>a} =0 uniformly in n. In a~co order to make the result valid it is necessary to enlarge the probability space so that there is a random variable with uniform distribution on [0, 11 which is independent of the process and also to properly define the notion of convergence. The proper notion of convergence is easy to state and it is that the subsequence { T(n(k))} converges to T as far as the process is concerned provided that the distributions of {T(n(k))} and of the process stopped at T(n(k)) converge in the usual way, i.e. weakly, to the distributions of T and of the process stopped at T. The enlargement of the space which is required is essential since it is easy to give examples which show that the result is false without it. This enlargement gives rise to randomized stopping times, so that the closure of the space of stopping times is the spaces of randomized stopping times. A result of independent interest obtained in the course of the proof is that if {T(n)} and {U(n)} are sequences of stopping times which do not grow too rapidly in the sense already given and if T(n)-U(n) converges to zero in probability then f(Xr(n), Xv(,)) converges to zero in probability as well, where X~ is the process and where f(x, y) is a function which can be taken to be the distance between x and y, when one exists, although in general the state space is not assumed to be a metric space. An overall assumption made about the process is that it is standard, where the definition of the standard process is given in the same way as in the Markov case. It is not necessary, however, to assume that the process is a Markov process. Many probabilistic constructions involve a passage from the discrete to the continuous. The main theorem in the paper gives a general method for proving
Frontiers in Digital Health, 2022
Identifying which patients are at higher risks of dying or being re-admitted often happens to be ... more Identifying which patients are at higher risks of dying or being re-admitted often happens to be resource- and life- saving, thus is a very important and challenging task for healthcare text analytics. While many successful approaches exist to predict such clinical events based on categorical and numerical variables, a large amount of health records exists in the format of raw text such as clinical notes or discharge summaries. However, the text-analytics models applied to free-form natural language found in those notes are lagging behind the break-throughs happening in the other domains and remain to be primarily based on older bag-of-words technologies. As a result, they rarely reach the accuracy level acceptable for the clinicians. In spite of their success in other domains, the superiority of deep neural approaches over classical bags of words for this task has not yet been convincingly demonstrated. Also, while some successful experiments have been reported, the most recent bre...