A continuous-speech interface to a decision support system: I. Techniques to accommodate for misrecognized input.

OBJECTIVE: Develop a continuous-speech interface that allows flexible input of clinical findings into a medical diagnostic application. DESIGN: The authors' program allows users to enter clinical findings using their own vernacular. It displays from the diagnostic program's controlled vocabulary a list of terms that most closely matches the input, and allows the user to select the single best term. The interface program includes two components: a speech-recognition component that converts utterances into text strings, and a language-processing component that matches recognized text strings with controlled-vocabulary terms. The speech-recognition component is composed of commercially available speech-recognition hardware and software, and developer-created grammars, which specify the language to be recognized. The language-processing component is composed of a translator, which extracts a canonical form from both recognized text strings and controlled-vocabulary terms, and a matcher, which measures the similarity between the two canonical forms. RESULTS: The authors discovered that grammars constructed by a physician, who could anticipate how users might speak findings, supported speech recognition better than did grammars constructed programmatically from the controlled vocabulary. However, this programmatic method of grammar construction was more time efficient and better supported long-term maintenance of the grammars. The authors also found that language-processing techniques recovered some of the information lost due to speech misrecognition, but were dependent on the completeness of supporting synonym dictionaries. CONCLUSIONS: The authors' program demonstrated the feasibility of using continuous speech to enter findings into a medical application. However, improvements in speech-recognition technology and language-processing techniques are needed before natural continuous speech becomes an acceptable input modality for clinical applications.

A continuous-speech interface to a decision support system: II. An evaluation using a Wizard-of-Oz experimental paradigm.

OBJECTIVE: Evaluate the performance of a continuous-speech interface to a decision support system. DESIGN: The authors performed a prospective evaluation of a speech interface that matches unconstrained utterances of physicians with controlled-vocabulary terms from Quick Medical Reference (QMR). The performance of the speech interface was assessed in two stages: in the real-time experiment, physician subjects viewed audiovisual stimuli intended to evoke clinical findings, spoke a description of each finding into the speech interface, and then chose from a list generated by the interface the QMR term that most closely matched the finding. Subjects believed that the speech recognizer decoded their utterances; in reality, a hidden experimenter typed utterances into the interface (Wizard-of-Oz experimental design). Later, the authors replayed the same utterances through the speech recognizer and measured how accurately utterances matched with appropriate QMR terms using the results of the real-time experiment as the "gold standard." MEASUREMENTS: The authors measured how accurately the speech-recognition system converted input utterances to text strings (recognition accuracy) and how accurately the speech interface matched input utterances to appropriate QMR terms (semantic accuracy). RESULTS: Overall recognition accuracy was less than 50%. However, using language-processing techniques that match keywords in recognized utterances to keywords in QMR terms, the semantic accuracy of the system was 81%. CONCLUSIONS: Reasonable semantic accuracy was attained when language-processing techniques were used to accommodate for speech misrecognition. In addition, the Wizard-of-Oz experimental design offered many advantages for this evaluation. The authors believe that this technique may be useful to future evaluators of speech-input systems.

Graphical access to medical expert systems: IV. Experiments to determine the role of spoken input.

The goal of our research is to design improved interfaces for medical expert systems. Previously, the use of graphical techniques was explored to improve the acceptance by clinicians of the user interface. Now that devices that accept spoken input are available, we wish to design interfaces that take advantage of this potentially more natural modality for interaction. To understand how clinicians might want to speak to a medical decision-support system, we carried out an experiment that simulated the availability of a spoken interface to the ONCOCIN medical expert system. ONCOCIN provides therapy advice for patients on complex cancer therapy protocols based on a description of the patient's current medical status and laboratory-test values. In the experiment, we had oncologists present a clinical case while observing the ONCOCIN flowsheet display. A project member listened to the presentation and filled in values for the flowsheet, as well as introducing purposeful misunderstandings of the input. The results suggest that each individual developed a stereotypical grammar for communicating with the program. Our experience with the purposeful miscommunications suggests particular ways to tailor requests for repetition based on the part of the utterance that was not understood.

Graphical access to medical expert systems: V. Integration with continuous-speech recognition.

This paper describes three prototypes of computer-based clinical record-keeping tools that use a combination of window-based graphics and continuous speech in their user interfaces. Although many of today's commercial speech-recognition products achieve high rates of accuracy for large grammars (vocabularies of words or collections of sentences and phrases), they can only "listen for" (and therefore recognize) a limited number of words or phrases at a time. When a speech application requires a grammar whose size exceeds a speech-recognition product's limits, the application designer must partition the large grammar into several smaller ones and develop control mechanisms that permit users to select the grammar that contains the words or phrases they wish to utter. Furthermore, the user interfaces they design must provide feedback mechanisms that show users the scope of the selected grammars. The three prototypes described were designed to explore the use of window-based graphics as control and feedback mechanisms for continuous-speech recognition in medical applications. Our experiments indicate that window-based graphics can be effectively used to provide control and feedback for certain classes of speech applications, but they suggest that the techniques we describe will not suffice for applications whose grammars are very complex.

Sudden unexpected death in young adult due to right ventricular dysplasia.

This case report illustrates a rare familial cardiomyopathy first reported in the medical literature in 1982 known as right ventricular dysplasia (right ventricular cardiomyopathy). The patient is a young woman with a history of cardiac arrhythmias suspected to be associated with prolapsed mitral valve who presented to the Berks Country Coroner's office as a sudden unexpected death in a young adult. It is important to recognize the illustrated classic cardiac pathology of this rare entity for clinical management, as an anatomic explanation of cause of sudden death and for the accumulation of statistics to establish frequency, conditions of predisposition, response to therapy and predicted outcome.

Temporal arteritis presenting as ataxia and dementia.

The case reported here illustrates some of the protean manifestations of temporal arteritis. Perhaps, as more cases with atypical manifestations are described, physicians will become more alert to the possibility of this diagnosis. It is now our policy to include temporal arteritis high on the list of differential diagnoses for any neuropsychiatric, visual, or systemic complaint in an elderly patient, even in the absence of typical manifestations. A temporal artery biopsy done relatively early in undiagnosed illness may reveal a very treatable cause. We wish to emphasize the need for early consideration of temporal arteritis in the elderly patient with an elevated ESR and any unexplained neuropsychiatric problem.

Translation of Xenopus liver messenger RNA in Xenopus oocytes: vitellogenin synthesis and conversion to yolk platelet proteins.

Xenopus liver vitellogenin and albumin mRNAs injected into Xenopus oocytes are correctly translated, as shown by specific immunoprecipitation and co-electrophoresis with purified Xenopus vitellogenin (molecular weight 210,000 daltons) and albumin (molecular weight 72,000 daltons). Vitellogenin made in oocytes under the direction of injected liver mRNA is unstable compared to other proteins made on injected messengers (such as albumin and globin) and endogenous oocyte proteins (including actin), the half-life of newly made vitellogenin being about 8 hr. Pulse-chase experiments with 35S-methionine show vitellogenin to be a precursor to yolk platelet lipovitellin (molecular weight 120,000 daltons), while 3H-serine labeling demonstrates conversion to phosvitin (molecular weight 34,000 daltons). In contrast, injected 3H-serine 35S-methionine-labeled Xenopus vitellogenin protein is not converted to yolk platelet proteins and is degraded rather slowly (half-life, 23, 29 hr). Phosphorylation of serine residues in phosvitin can be detected in oocytes injected with 32PO4 or gamma-32P-ATP; thus exogenously derived yolk platelet protein is further modified, or turned over, once it is within the oocyte. Moreover, vitellogenin made in oocytes programed with liver mRNA is phosphorylated. Thus phosphorylation, assembly into yolk platelets, and cleavage are events that do not require vitellogenin supplied by the normal pathoways involved in yolk formation (synthesis and post-translational modification in the liver, transport in the serum, and follicle cell-dependent pinocytosis). Vitellogenin mRNA sediments at about 29S in a sucrose-SDS gradient, while albumin messenger peaks at 16S; both species contain poly(A). These liver mRNAs are functionally stable in oocytes for at least 5 days. Vitellogenin-forming activity, relative to albumin, actin, or total endogenous activity, increases with time, and the final rate of 2-2.5 times the initial rate is only reached 3 days after injection. The potentiation effect probably stems from an increase in the efficiency of translation of vitellogenin mRNA. The availability of homologous mRNAs now permits injected messenger to be used as a valide probe of oocyte function; the biological activity of mRNA from a non-ovarian Xenopus tissue proves that some at least of the translational systems within the Xenopus oocyte are not cell type-specific. Moreover, the whole cell system is eminently suitable for assaying putative translational (and possibly transcriptional) control elements from frog liver.

Signal sequences, secondary modification and the turnover of miscompartmentalized secretory proteins in Xenopus oocytes.

The cytoplasm of the Xenopus oocyte can be altered by the microinjection of proteins and the regulatory responses to such perturbations can then be studied. We have investigated proteolytic systems within the oocyte which may be involved in the maintenance of the integrity of the different subcellular compartments. Thus primary translation products, made in the wheat germ system under the direction of frog liver, chicken oviduct, rat liver rapidly sedimenting endoplasmic reticulum, rat seminal vesicle, guinea pig mammary gland or honey been venom gland RNA, were injected into oocytes. Their stability in the frog cell cytosol was in general low compared to that of their processed counterparts. The latter were usually obtained by collecting the heterologous proteins exported by RNA-injected oocytes. Electrophoretic analysis of oocytes injected with particular primary and processed polypeptides permitted measurement of the stabilities of proteins differing only by the presence or absence of a detachable signal sequence, or by the presence of a specific secondary modification. The effect of the latter on protein stability appears slight. However, the presence of a detachable signal sequence destabilizes those miscompartmentalized secretory proteins which are otherwise stable. Indeed all other results are consistent with this concept for they show that primary translation products are in general much less and are never more stable than their processed counterparts. Thus we provide evidence that errors of compartmentation can be corrected in living cells and that this process is often facilitated by the properties conferred on a protein by a detachable signal sequence.

The sequestration, processing and retention of honey-bee promelittin made in amphibian oocytes.

Messenger RNAs from one kind of secretory cell can be introduced into the cytoplasm of another: the heterologous proteins formed by the recipient cell are usually processed and topologically segregated in the manner characteristic of the donor cell. Xenopus oocytes injected with honey-bee venom gland RNA provide some support for this generalization, but also reveal important exceptions to it. Thus, the frog cell makes a small polypeptide whose partial sequence matches perfectly that of insect promelittin, except that the product formed in oocytes ends at the C terminus with a glycine as opposed to a glutamine amide residue. N-terminal heterogeneity is seen in protoxin made in oocytes and venom gland cells, and species shorter by two residues are seen in both tissues. We suggest that the oocyte contains a dipeptidylpeptidase. Promelittin made in oocytes is barely detectable in the cytosol but is found associated with a vesicle fraction which also contains some newly synthesized endogenous oocyte proteins. The association with vesicles is long-lasting; thus promelittin is retained slightly more efficiently than sequestered oocyte proteins, and an incubation period of about two weeks is required to reduce by half the amount of these endogenous vesicle proteins. Thus neither promelittin nor any products derived from it are secreted rapidly. Gel analysis fails to reveal promelittin in the medium surrounding the oocyte, although traces can be detected by assaying for a characteristic heptapeptide. Such small amounts could result from slow secretion or leakage. Melittin could not be detected by gel analysis or peptide assay. The retention of the honey-bee protein within the frog cell is discussed in terms of the specificity of the processing systems and secretory pathways of venom gland cells and oocytes. We suggest that whilst some export mechanisms function efficiently in a wide variety of cells, others do not, and may even be restricted to specific cell types.

Synthesis and insertion, both in vivo and in vitro, of rat-liver cytochrome P-450 and epoxide hydratase into Xenopus laevis membranes.

We described whole cell and cell-free systems capable of inserting into membranes cytochrome P-450 and epoxide hydratase made under the direction of rat liver RNA. The systems have been used to study the pathways followed by newly made secretory and integral membrane proteins. The cell-free system contains Xenopus laevis embryo membranes, and demonstrates competition for a common receptor between cytochrome P-450 and epoxide hydratase, and normal secretory proteins: evidence is provided for differential membrane receptor affinity. Thus, synthesis of secretory and membrane proteins appears to involve a common initial pathway. Microinjection of rat liver RNA into whole oocytes suggests that membrane insertion is neither cell type nor species specific, because functional rat liver enzymes are found inserted in the endoplasmic reticulum of the frog cell. Nonetheless, insertion is highly selective since albumin and several other proteins made under the direction of the injected liver RNA are sequestered within membrane vesicles and are then secreted by the oocyte, whilst epoxide hydratase and cytochrome P-450 are inserted into membranes but are not secreted.

The integration of a continuous-speech-recognition system with the QMR diagnostic program.

We describe a continuous-speech interface for Quick Medical Reference (QMR), which allows physicians to input spoken descriptions of physical-examination findings, or observations. We analyze the difficulties in designing a continuous-speech interface for systems that use medical terminology. We present a method for matching spoken findings names expressed in natural language to QMR terms. The method is based on a semantic representation of findings that both minimize the effect of misrecognition and derive grammars that are necessary for supporting the recognition process.

The absence from the oocyte secretory apparatus of a protein kinase capable of phosphorylating sequestered caseins.

The lactating guinea-pig mammary gland synthesizes and secretes four major milk proteins, i.e., three caseins and alpha-lactalbumin. Of these, the caseins are highly phosphorylated, a post-translational event which in the mammary gland involves a specific casein kinase, which is an integral membrane protein probably of Golgi origin. The microinjection of milk protein mRNA into Xenopus oocytes in the presence of [35S]methionine leads to the synthesis, sequestration, and secretion of proteins which coelectrophorese with alpha-lactalbumin and with partially processed caseins. That the secreted caseins were not phosphorylated was shown by the use of 32P. Either the oocytes were injected with mammary gland mRNA followed by incubation with [32P]phosphate containing media or the mRNA was co-injected with [gamma-32P]ATP and the oocytes were then incubated. In neither case were 32P-labeled caseins secreted. Golgi-rich fractions, identified by the marker enzyme galactosyltransferase, were isolated from the postnuclear supernatant of both oocytes and lactating mammary gland by sucrose density gradient fractionation. In contrast to the mammary gland fractions those derived from the oocytes contained no detectable casein kinase activity. Homogenates of oocytes do effect the phosphorylation of casein but the enzyme activity appears to be present in the soluble fraction and is not membrane bound. It is concluded that the Xenopus oocyte lacks the specific kinase that in the mammary gland phosphorylates sequestered caseins and that the phosphorylation of the caseins is not a prerequisite for their secretion by the oocyte.