Sequence Prediction With Sparse Distributed Hyperdimensional Coding Applied to the Analysis of Mobile Phone Use Patterns.

Modeling and prediction of temporal sequences is central to many signal processing and machine learning applications. Prediction based on sequence history is typically performed using parametric models, such as fixed-order Markov chains ( n -grams), approximations of high-order Markov processes, such as mixed-order Markov models or mixtures of lagged bigram models, or with other machine learning techniques. This paper presents a method for sequence prediction based on sparse hyperdimensional coding of the sequence structure and describes how higher order temporal structures can be utilized in sparse coding in a balanced manner. The method is purely incremental, allowing real-time online learning and prediction with limited computational resources. Experiments with prediction of mobile phone use patterns, including the prediction of the next launched application, the next GPS location of the user, and the next artist played with the phone media player, reveal that the proposed method is able to capture the relevant variable-order structure from the sequences. In comparison with the n -grams and the mixed-order Markov models, the sparse hyperdimensional predictor clearly outperforms its peers in terms of unweighted average recall and achieves an equal level of weighted average recall as the mixed-order Markov chain but without the batch training of the mixed-order model.

EVAR may reduce the risk of aneurysm rupture despite persisting type Ia endoleaks.

PURPOSE: To determine the risk of aneurysm rupture in patients with persisting proximal type Ia endoleaks following endovascular aneurysm repair (EVAR) in comparison to the risk of rupture of untreated abdominal aortic aneurysms (AAA) of similar size. METHODS: Among 400 patients who where treated with EVAR from 1996 to 2003 at a single center, 21 (5.3%) patients (13 men; mean age 78.0±5.0 years, range 67-86) with large (≥5.5 cm) aneurysms had imaging evidence of type Ia endoleaks that persisted >10 months (type Ia group) despite secondary endovascular treatment. These patients were compared to 24 untreated AAA patients (17 men; mean age 73.8±5.2 years, range 64-88) with large aneurysms from a separate geographic region with a well-established aneurysm treatment program before EVAR became available (1990-1998). RESULTS: There were no significant differences between the type Ia and the untreated AAA patients with regard to age (79±8 vs. 74±5 years), gender (38% vs. 29% women), baseline aneurysm diameter (6.1±0.7 vs. 6.4±0.9 cm), or length of follow-up (32±23 vs. 29±40 months). During the follow-up period, the rate of aneurysm enlargement was significantly lower in type Ia patients (0.19 cm/y) than in untreated AAA patients (0.54 cm/y, p = 0.03). One (4.8%) patient with a persisting type Ia endoleak and 2-cm aneurysm enlargement (0.8 cm/y) had aneurysm rupture after 2.5 years, while 12 (50%) of the 24 untreated aneurysms ruptured (p = 0.001), which was the primary cause of death in this group. The rupture rate was 1.8 per 100 patient-years in the type Ia group and 20.7 per 100 patient-years in the untreated AAA group. Aneurysm-related mortality was significantly reduced in the type Ia group compared to the untreated AAA group at 36 months (11% vs. 52%, p = 0.004). In the multivariate analysis, factors associated with death were an untreated AAA (odds ratio 97, p = 0.004), female gender (odds ratio 9.7, p = 0.02), and baseline aneurysm size (odds ratio 4.7/cm, p = 0.03). CONCLUSION: This study suggests that EVAR may reduce the risk of rupture and aneurysm-related death despite the presence of a persisting type Ia endoleak. This finding is limited to patients with aortic endografts that are in good position. The mechanism of protection from rupture is unclear but may be related to reducing the rate of aneurysm enlargement.

Postthrombotic syndrome after isolated calf deep venous thrombosis: the role of popliteal reflux.

OBJECTIVE: The purpose of this study was to evaluate the long-term clinical and hemodynamic outcomes after isolated first-time calf deep venous thrombosis (cDVT). METHODS: This retrospective clinical study was set in an academic referral center. From 1990 to 1994, 617 patients were seen with acute DVT. This number included 82 patients with phlebographically confirmed cDVT. Of those patients, 50 attended the clinical assessment 6 to 10 years (mean, 8.4 years) after the acute event. All patients with cDVT underwent treatment with anticoagulant therapy (96% heparin and warfarin, 4% only warfarin). The duration of the heparin treatment was 4.0 to 8.0 days (mean, 6.4 days), and warfarin was given for 2.0 to 7.5 months (mean, 3.4 months). Compression stockings were used regularly (mean, 9.2 months; range, 0.25 to 64 months) in 30% of the patients after acute cDVT. The initial ipsilateral phlebograms were reevaluated to confirm the diagnosis of cDVT without popliteal involvement. The clinical assessment included evaluation of both legs according to CEAP clinical classification C0-6. Bilateral color-flow duplex scan imaging was performed to assess reflux in deep popliteal segments. Photoplethysmographic measurement of venous refilling time was conducted in both legs to observe deep reflux. RESULTS: The mean age was 57 years (range, 30 to 76 years) at the time of the clinical assessment. Cause of acute cDVT was idiopathy in 52%, coagulopathy in 2%, trauma in 10%, immobilization in 22%, and postoperative in 14% of the cases. During the follow-up period, seven recurrent DVTs (14%) were seen. In the clinical assessment, 17 legs (34%) with previous cDVT had skin changes (CEAP C4-6). No active ulcers were found. Contralaterally, the frequency of C4-6 was 10% (n = 5; P <.05). After exclusion of recurrent DVTs, the distribution of the clinical classification still remained the same. Deep popliteal reflux was detected in 20 legs (40%) with previous cDVT. Contralaterally, popliteal reflux was seen in nine cases (18%; P <.05). Plethysmography showed deep reflux in 16 legs (33%) with cDVT and in nine cases (18%) contralaterally (P >.05). A significant association was found between deep popliteal reflux and skin changes (P <.05). CONCLUSION: In the long-term follow-up, cDVT may lead to significant postthrombotic disease. Reflux in the primarily uninvolved popliteal vein is frequent and may be associated with more severe disease.

Regional differences in the use of a vascular surgical service and incidence of amputations in a well-defined geographical area.

OBJECTIVE: To evaluate regional differences in the use of a vascular surgical service in the treatment of critical lower limb ischaemia and incidence of amputation in the 1990s in a well-defined geographical area. DESIGN: Retrospective study. SETTING: One university and five county hospitals, Finland. SUBJECTS: All referrals to the university hospital vascular surgical unit for chronic critical lower limb ischaemia and the number of major amputations in the region. MAIN OUTCOME MEASURES: Numbers of new vascular surgical consultations and amputations in 11 municipalities. Correlation between numbers of consultations and amputations. RESULTS: Between the subregions the age-standardised incidence of new vascular surgical consultations in the 15-85 year old population varied from 52.4 to 104.7/10(5) and the incidence of amputation from 10.2 to 24.8/10(5). There was an inverse correlation between the numbers of consultations and amputations. The most significant inverse correlation was between consultations and below knee amputations in diabetic patients (r = -0.70). For above knee amputations there was no correlation (r = -0.21). CONCLUSION: An active referral policy leads to reduced amputation rates.